VY_Project4_Modeling.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Project 4: Web Scraping Job Postings\n",
    "\n",
    "## Business Case Overview\n",
    "\n",
    "You're working as a data scientist for a contracting firm that's rapidly expanding. Now that they have their most valuable employee (you!), they need to leverage data to win more contracts. Your firm offers technology and scientific solutions and wants to be competitive in the hiring market. Your principal has two main objectives:\n",
    "\n",
    "   1. Determine the industry factors that are most important in predicting the salary amounts for these data.\n",
    "   2. Determine the factors that distinguish job categories and titles from each other. For example, can required skills accurately predict job title?\n",
    "\n",
    "To limit the scope, your principal has suggested that you *focus on data-related job postings*, e.g. data scientist, data analyst, research scientist, business intelligence, and any others you might think of. You may also want to decrease the scope by *limiting your search to a single region.*\n",
    "\n",
    "Hint: Aggregators like [Indeed.com](https://www.indeed.com) regularly pool job postings from a variety of markets and industries. \n",
    "\n",
    "**Goal:** Scrape your own data from a job aggregation tool like Indeed.com in order to collect the data to best answer these two questions.\n",
    "\n",
    "---\n",
    "\n",
    "## Directions\n",
    "\n",
    "In this project you will be leveraging a variety of skills. The first will be to use the web-scraping and/or API techniques you've learned to collect data on data jobs from Indeed.com or another aggregator. Once you have collected and cleaned the data, you will use it to answer the two questions described above.\n",
    "\n",
    "### QUESTION 1: Factors that impact salary\n",
    "\n",
    "To predict salary you will be building either a classification or regression model, using features like the location, title, and summary of the job. If framing this as a regression problem, you will be estimating the listed salary amounts. You may instead choose to frame this as a classification problem, in which case you will create labels from these salaries (high vs. low salary, for example) according to thresholds (such as median salary).\n",
    "\n",
    "You have learned a variety of new skills and models that may be useful for this problem:\n",
    "- NLP\n",
    "- Unsupervised learning and dimensionality reduction techniques (PCA, clustering)\n",
    "- Ensemble methods and decision tree models\n",
    "- SVM models\n",
    "\n",
    "Whatever you decide to use, the most important thing is to justify your choices and interpret your results. *Communication of your process is key.* Note that most listings **DO NOT** come with salary information. You'll need to able to extrapolate or predict the expected salaries for these listings.\n",
    "\n",
    "### QUESTION 2: Factors that distinguish job category\n",
    "\n",
    "Using the job postings you scraped for part 1 (or potentially new job postings from a second round of scraping), identify features in the data related to job postings that can distinguish job titles from each other. There are a variety of interesting ways you can frame the target variable, for example:\n",
    "- What components of a job posting distinguish data scientists from other data jobs?\n",
    "- What features are important for distinguishing junior vs. senior positions?\n",
    "- Do the requirements for titles vary significantly with industry (e.g. healthcare vs. government)?\n",
    "\n",
    "You may end up making multiple classification models to tackle different questions. Be sure to clearly explain your hypotheses and framing, any feature engineering, and what your target variables are. The type of classification model you choose is up to you. Be sure to interpret your results and evaluate your models' performance.\n",
    "\n",
    "\n",
    "### BONUS PROBLEM\n",
    "\n",
    "Your boss would rather tell a client incorrectly that they would get a lower salary job than tell a client incorrectly that they would get a high salary job. Adjust one of your models to ease his mind, and explain what it is doing and any tradeoffs. Plot the ROC curve.\n",
    "\n",
    "---\n",
    "\n",
    "## Requirements\n",
    "\n",
    "1. Scrape and prepare your own data.\n",
    "\n",
    "2. **Create and compare at least two models for each section**. One of the two models should be a decision tree or ensemble model. The other can be a classifier or regression of your choosing (e.g. Ridge, logistic regression, KNN, SVM, etc).\n",
    "   - Section 1: Job Salary Trends\n",
    "   - Section 2: Job Category Factors\n",
    "\n",
    "3. Prepare a polished Jupyter Notebook with your analysis for a peer audience of data scientists. \n",
    "   - Make sure to clearly describe and label each section.\n",
    "   - Comment on your code so that others could, in theory, replicate your work.\n",
    "\n",
    "4. A brief writeup in an executive summary, written for a non-technical audience.\n",
    "   - Writeups should be at least 500-1000 words, defining any technical terms, explaining your approach, as well as any risks and limitations.\n",
    "\n",
    "#### BONUS\n",
    "\n",
    "5. Answer the salary discussion by using your model to explain the tradeoffs between detecting high vs low salary positions.\n",
    "\n",
    "6. Convert your executive summary into a public blog post of at least 500 words, in which you document your approach in a tutorial for other aspiring data scientists. Link to this in your notebook.\n",
    "\n",
    "---\n",
    "\n",
    "## Suggestions for Getting Started\n",
    "\n",
    "1. Collect data from [Indeed.com](www.indeed.com) (or another aggregator) on data-related jobs to use in predicting salary trends for your analysis.\n",
    "  - Select and parse data from *at least 1000 postings* for jobs, potentially from multiple location searches.\n",
    "2. Find out what factors most directly impact salaries (e.g. title, location, department, etc).\n",
    "  - Test, validate, and describe your models. What factors predict salary category? How do your models perform?\n",
    "3. Discover which features have the greatest importance when determining a low vs. high paying job.\n",
    "  - Your Boss is interested in what overall features hold the greatest significance.\n",
    "  - HR is interested in which SKILLS and KEY WORDS hold the greatest significance.   \n",
    "4. Author an executive summary that details the highlights of your analysis for a non-technical audience.\n",
    "5. If tackling the bonus question, try framing the salary problem as a classification problem detecting low vs. high salary positions.\n",
    "\n",
    "---"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Basic Imports\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "dataanalystdf = pd.read_csv(\"Data_Analyst_MCF.csv\", index_col=0)\n",
    "dataengineerdf = pd.read_csv(\"Data_Engineer_MCF.csv\", index_col=0)\n",
    "dataanalyticsdf = pd.read_csv(\"Data_Analytics_MCF.csv\", index_col=0)\n",
    "datascientistdf = pd.read_csv(\"Data_Scientist_MCF.csv\", index_col=0)\n",
    "datasciencedf = pd.read_csv(\"Data_Science_MCF.csv\", index_col=0)\n",
    "datawarehousedf = pd.read_csv(\"DWH_MCF.csv\", index_col=0)\n",
    "databasedf = pd.read_csv(\"DB_MCF.csv\", index_col=0)\n",
    "datamanagerdf = pd.read_csv(\"Data_Manager_MCF.csv\", index_col=0)\n",
    "analyticsdf = pd.read_csv(\"Analytics_MCF.csv\", index_col=0 ) \n",
    "architectdf = pd.read_csv(\"Data_Architect_MCF.csv\", index_col=0)\n",
    "etldf = pd.read_csv(\"ETL_MCF.csv\", index_col=0)\n",
    "businessintelligencedf = pd.read_csv(\"BI_MCF.csv\", index_col=0 )\n",
    "bigdatadf = pd.read_csv(\"Big_Data_MCF.csv\", index_col=0)                                     \n",
    "artificialintelligencedf = pd.read_csv(\"AI_MCF.csv\", index_col=0)\n",
    "deeplearndf = pd.read_csv(\"Deep_Learning_MCF.csv\", index_col=0)\n",
    "machinelearndf = pd.read_csv(\"ML_MCF.csv\", index_col=0)\n",
    "nlpdf = pd.read_csv(\"NLP_MCF.csv\", index_col=0)\n",
    "researchscientistdf = pd.read_csv(\"Research_Scientist_MCF.csv\", index_col=0)\n",
    "\n",
    "jobs = pd.concat([ dataanalystdf, dataengineerdf, dataanalyticsdf, datascientistdf, datasciencedf, datawarehousedf, \\\n",
    "                  databasedf, datamanagerdf, analyticsdf, architectdf, etldf, businessintelligencedf, bigdatadf, \\\n",
    "                  artificialintelligencedf, deeplearndf, machinelearndf, nlpdf, researchscientistdf])\n",
    "jobs.to_csv(\"mycareersfuture.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_csv(\"mycareersfuture.csv\", index_col=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(3094, 18)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Dropping rows with either blank or 0 salary information\n",
    "df = df.drop(index=df[df[\"Min Salary\"].isna()].index).reset_index(drop=True)\n",
    "df = df.drop(index=df[df[\"Min Salary\"] == 0].index).reset_index(drop=True)\n",
    "\n",
    "# Dropping duplicates with exact same Job ID from web scraping\n",
    "df = df.drop_duplicates([\"Job ID\"]).reset_index(drop=True)\n",
    "df = df.drop([\"Job ID\", \"URL\"], axis=1)\n",
    "\n",
    "# Dropping duplicates which have same Job Title from same Company and Job Type\n",
    "df = df.drop_duplicates([\"Job Title\", \"Job Type\", \"Company Name\"]).reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1384"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Total number of records remaining after removing duplicates and blank rows\n",
    "len(df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Removing special characters in every text in Job Description\n",
    "specialchar = [\"\\xa0\", \"\\uf0b7\", \"\\t\", \"· \", \"\\r\\n\"]\n",
    "for c in specialchar:\n",
    "    df[\"Job Description\"] = df[\"Job Description\"].map(lambda x: x.replace(c, \"\").lower())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Creating Mean Salary column as the aggregated figure to classify\n",
    "df[\"Mean Salary\"] = (df[\"Max Salary\"] + df[\"Min Salary\"])/2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Data Scientist                                                  33\n",
       "Data Engineer                                                   24\n",
       "Software Engineer                                               11\n",
       "Data Analyst                                                    10\n",
       "Business Analyst                                                 9\n",
       "Database Administrator                                           6\n",
       "Full Stack Developer                                             6\n",
       "Software Developer                                               6\n",
       "Human Resource Executive                                         6\n",
       "Lead Data Scientist                                              5\n",
       "Manager                                                          5\n",
       "Marketing Executive                                              5\n",
       "Senior Software Developer                                        5\n",
       "Operations Executive                                             4\n",
       "Technical Consultant                                             4\n",
       "Business Development Manager                                     4\n",
       "Sales Manager                                                    4\n",
       "Recruitment Consultant                                           4\n",
       "Research Fellow                                                  4\n",
       "Research Associate                                               4\n",
       "Research Engineer                                                4\n",
       "Senior Data Analyst                                              4\n",
       "Project Manager                                                  4\n",
       "ETL Developer                                                    3\n",
       "Digital Marketing Executive                                      3\n",
       "Business Development Executive                                   3\n",
       "Research Assistant                                               3\n",
       "Systems Engineer                                                 3\n",
       "Information Technology Consultant                                3\n",
       "Senior Manager                                                   3\n",
       "                                                                ..\n",
       "Marketing Coordinator                                            1\n",
       "Cyber Operations Analyst L1                                      1\n",
       "Dev Ops Engineer                                                 1\n",
       "Assistant Engineer                                               1\n",
       "Senior Consultant - Finance Transformation                       1\n",
       "Frontend UI / UX Developer                                       1\n",
       "REGIONAL SALES ACCOUNT EXECUTIVES                                1\n",
       "Scrum Master                                                     1\n",
       "UI Web Developers                                                1\n",
       "IT Chatbot Engineer                                              1\n",
       "Corporate Finance Researcher and Translator                      1\n",
       "Assistant Manager, GeoAnalytics                                  1\n",
       "Senior Analyst, Digital Solution Development                     1\n",
       "Data Specialist (Japanese Language proficient)                   1\n",
       "Regional Product Manager                                         1\n",
       "Cyber Investigations and Forensic investigation                  1\n",
       "Big Data Architect                                               1\n",
       "Dot Net Developer                                                1\n",
       "Senior  /  R&D Engineer (AI, Deep Learning, Computer Vision)     1\n",
       "Sales Executive (Corporate Sales)                                1\n",
       "software Engineer                                                1\n",
       "Sales & Marketing Executive Chemical                             1\n",
       "Deployment Specialist (Application Infrastructure support)       1\n",
       "Recruitment Manager                                              1\n",
       "Siebel CRM developer (EAI / Workflow)                            1\n",
       "UI, UX Designer                                                  1\n",
       "Senior Data Research Engineer                                    1\n",
       "Corporate Secretarial Executive                                  1\n",
       "Executive Assistant                                              1\n",
       "Operations Manager                                               1\n",
       "Name: Job Title, Length: 1148, dtype: int64"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Checking if there are any duplicated unique Job Titles\n",
    "df[\"Job Title\"].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Salary > 40k a month is quite unlikely realistically, to recognize as Annual salary instead\n",
    "df.loc[df[df[\"Mean Salary\"] > 40000].index, \"Salary Paid\"] = \"Annually\"\n",
    "df.loc[df[df[\"Salary Paid\"] == \"Annually\"].index, \"Mean Salary\"] = df.loc[df[df[\"Salary Paid\"] == \"Annually\"].index, \"Mean Salary\"].apply(lambda x: x/12)\n",
    "df.loc[df[df[\"Salary Paid\"] == \"Annually\"].index, \"Salary Paid\"] = \"Monthly\"\n",
    "# Salary Paid unknown, imputed as Monthly\n",
    "df.loc[df[df[\"Salary Paid\"].isna()].index, \"Salary Paid\"] = \"Monthly\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZAAAAEWCAYAAABIVsEJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvOIA7rQAAIABJREFUeJzt3Xd8XNWZ+P/Po1G1ZEtWcS+ysSmmgzAQklCz2KGYJLDBTrJmcUJ+G9gUSBZ7ybJZEr5AEtaEBZIFTCAsxhCqIQYSahKKG8UVgVwlV9kqliVZ9fn9cY/MzPWMNJI1mhnpeb9e89J9zj333HNH0jxz77lFVBVjjDGmu1Li3QFjjDHJyRKIMcaYHrEEYowxpkcsgRhjjOkRSyDGGGN6xBKIMcaYHrEEYkwMiMhwEfmriNSJyJ3x7k+siYiKyKQeLjtORPaLSKC3+2ViyxKI6RYR2Swije4fvuM1Kt79SkDXAHuAIap6g3+miDzsPnQv9ZXf5cqv6qN+Bq97hoh8KCL7RGSPiLwmIsWxXq+qblXVHFVti/W6TO+yBGJ64hL3D9/x2u6vICKp8ehYAhkPrNPOr9T9BJjdEbj37ApgQ4z7dgi39/AH4AYgF5gA3Ae0x3i9A/3vJKlZAjG9QkSK3TfnOSKyFXjdlZ8hIu+ISI2IfCQi5wQtM0FE3nKHef4iIveIyP+5eeeISIVvHZtF5AI3nSIic0Vkg4jsFZEnRSTf15fZIrLVfZu+KaidgIj8u1u2TkRWishYEbnXf7hJRF4QkR9G2ObPichyEal1Pz/nyh/GSwz/5vbQLojwtr0AnCUiQ108DVgF7PSt52oRWS8i1SLyioiMD5r3GxEpd3sNK0XkC0Hzfubelz+47VwrIiUR+nISsElVX1NPnao+rapbXVtTReRd93vc4X5X6RHel4tE5APXp3IR+VnQvEP+ToLKUl2dXBFZ4NazTUR+0XF4S0Qmub+ZWvd7fSLC9pg+YAnE9LazgWOAC0VkNPAn4BdAPvBj4GkRKXJ1FwIrgULg5wR9G4/C94HL3PpGAdXAvb46nweOAs4HbhaRY1z59cBM4MvAEOBqoAF4BJgpIikAIlLoln3cv3KXrP4E3A0UAP8N/ElEClT1KuAx4JduD+3VCNtwAFgMXOnif8LbCwhez2XAvwNfBYqAv/n6sxzvwz8f7/38o4hkBs2/FFgE5Ll13ROhL+8DR4vIfBE5V0RyfPPbgB/h/a7OxHtfvhehrXq3LXnARcC/uO0IdvDvJMzyjwCtwCTgZOAfgG+7eT8H/gwMBcYA/xOhD6YvqKq97BX1C9gM7Adq3Os5V14MKDAxqO6NwKO+5V/BSxTj8D4ksoPmLQT+z02fA1SEWfcFbno9cH7QvJFAC5Aa1JcxQfOXAVe66VJgRoTtWw98yU1fByyJUO9bwDJf2bvAVW76YeAXnbyPD+Ml1s+75XKBXUAW8Pegdl4C5gQtl4KX7MZHaLcaONFN/wx4NWjeFKCxkz6dATwJVOIlt4eBnAh1fwg8GxQrMClC3buA+Z38nXSUpQLDgSYgK2j+TOANN/0H4P7g36294veyPRDTE5epap57+b9ZlgdNjweucIc9akSkBu8DcyRur0FV64Pqb+lGH8YDzwa1ux7vW/LwoDrBh4IagI5v1WOJPM7wCPBNN/1N4NEI9UaF6e8WYHRUvXdU9e94exY/BV5U1UZflfHAb4K2swqQjvWIyA3u8Fatm5+Lt5fQwf8eZEYad1DV91T1H1W1CPgC8EXgJreeI0XkRRHZKSL7gP/nW89BInK6iLwhIpUiUgv8f2HqlodZtGN704AdQdv8v8AwN//f3PYvc4fkro7QjukDlkBMbwseNC7H2wPJC3plq+rtwA5gqIhkB9UfFzRdDwzqCNwx8KKg+eXAdF/bmaq6LYo+lgNHRJj3f8AMETkR7xDLcxHqbcf7sAs2Dohm/eHWeQO+w1dOOfBd33Zmqeo7brzjRuAfgaGqmgfU4n3AHhZVXQ48Axznin4LfAxMVtUheIfVIq1nId7hsrGqmgv8LkzdSCcXlOPtgRQGbe8QVT3W9Wunqn5HVUcB3wXukx6ePmwOnyUQE0v/B1wiIhe6getMNzg+RlW3ACuA/xKRdBH5PHBJ0LKf4H1bvkhE0vC+oWcEzf8dcGvHgLKIFInIjCj79SDwcxGZLJ4TRKQAQFUr8MYVHgWeDrNH0GEJcKSIzBKRVBH5Ot4hohej7EOwu4EvAX8NM+93wDwRORYODjBf4eYNxjsMWAmkisjNeGM63SYinxeR74jIMBcfjTd+8l7QuvYB+928f+mkucFAlaoeEJGpwKxo+6GqO/DGOO4UkSHinSxxhIic7fp1hYiMcdWr8RKRnf4bJ5ZATMyoajkwA+/baiXet8uf8Nnf3SzgdLzDMv9J0DdwVa3FG6R9EO9bfT0QfFbWb/C+5f5ZROrwPuhOj7Jr/413rP/PeB+KC/DGHjo8AhxP5MNXqOpe4GK8PYe9eIdWLlbVPVH2IbitKnVnP4WZ9yxwB7DIHTpaA0x3s1/BGyP5BO/w2QEiHxrqSg1ewlgtIvuBl4FngV+6+T/G+33VAQ8AnZ399D3gFvd7uRnvve6OfwLSgXV4SeIpvMOeAKcBS10fFwM/UNVN3Wzf9BIJ8zdrTFy40z0nqeo3u6ob4358EW/vqVhVY3odhDHJzPZAjAniDpf9AHjQkocxnbMEYozjrhOpwTtcclecu2NMwrNDWMYYY3rE9kCMMcb0SL++kVlhYaEWFxfHuxtJZ/v27YwaNSq4AEbZDXeNGShWrly5x11Q2ql+nUCKi4tZsWJFvLuRdFauXMmpp54aXADBsTGmXxORqO4KYYewjDHG9IglEHOIkpISf0F8OmKMSWiWQIwxxvSIJRBjjDE9YgnEHOI///M//QXx6YgxJqH16wsJS0pK1M7CMsaY7hGRlara5eCn7YGYQ4zyX/Nh14AYY8KIKoGIyDQRKRWRMhGZG2Z+hog84eYvFZHioHnzXHmpiFzYVZsicp0rU/dM6uD1nCMiH7onkb3Vkw02XduxY4e/ID4dMcYktC4TiHsS3L14zyCYAswUkSm+anPwHk86CZiP9/wCXL0rgWOBaXhPDwt00ebbwAX4HhcqInnAfcCl7ulkV2CMMSZuorkSfSpQpqobAURkEd5DgtYF1ZkB/MxNPwXcIyLiyhepahOwSUTKXHtEalNVP3Bl/n7MAp5R1a0Aqrq7G9tpuuGUU045OL1w6VamHXUcLy/derBs1unjwi1mjBlgojmENZrQp5xVuLKwdVS1Fe+5zAWdLBtNm35H4j1D+00RWSki/xSukohcIyIrRGRFZWVlF02acFauXBkSv/zIn+LUE2NMIosmgRyyK4D3HOJo6nS3vDOpwKnARcCFwH+IyJGHNKJ6v6qWqGpJUVGX9wIzYVxzzTUh8dTbDhn2MsaYqBJIBTA2KB4DbI9UR0RSgVy851xHWjaaNsP142VVrXfPnf4rcGIU/Tfd9MADD4TEk55/PE49McYksmjGQJYDk0VkArANb1B8lq/OYmA28C5wOfC6qqqILAYWish/A6OAycAyvD2Qrtr0ex5vbCUVSAdOxxuwN71kYdA4R/C0McaE02UCUdVWEbkOeAUIAA+p6loRuQVYoaqLgQXAo26QvAovIeDqPYk34N4KXKuqbeCdrutv05V/H/g3YASwSkSWqOq3VXW9iLwMrALa8Z5Zvab33gpjjDHdYVeim4M69jqqK3cxtGj4wfKsyl00BsV2FpYx/ZtdiW56bNPHq0PifF9sjDFgCcSEcedP5oTEZ/tiY4wBSyDGGGN6yBKIMcaYHrEEYg4xZ+5tIfFSX2yMMWAJxIRx3mWhl+RsuKyrS3SMMQORJRBziG+cMT4knuWLjTEGLIEYY4zpIUsgxhhjesQSiDnEyWedHxJv88XGGAOWQEwYP77zoZD4LV9sjDFgCcSE8esbrg6Jz/bFxhgDlkBMGB+8/VpIPNoXG2MMWAIxxhjTQ5ZAjDHG9IglEHOIx97bEhIv9MXGGANRJhARmSYipSJSJiJzw8zPEJEn3PylIlIcNG+eKy8VkQu7alNErnNlKiKFYdZ1moi0icjl3d1YE53Xn1sYEh/hi40xBqJIICISAO4FpgNTgJkiMsVXbQ5QraqT8J5Tfodbdgre422PBaYB94lIoIs23wYuAA752uuWuwPvUbgmRhbcPi8kPt0XG2MMRLcHMhUoU9WNqtoMLAJm+OrMAB5x008B54uIuPJFqtqkqpuAMtdexDZV9QNV3RyhL/8KPA3sjnYDjTHGxEY0CWQ0UB4UV7iysHVUtRWoBQo6WTaaNkOIyGjgK8DvouizMcaYGIsmgUiYMo2yTnfLO3MXcKOqtnVWSUSuEZEVIrKisrKyiyZNODf8akFI/JYvNsYYgNQo6lQAY4PiMcD2CHUqRCQVyAWquli2qzb9SoBF3pExCoEvi0irqj4XXElV7wfuBygpKekqKZkwJhx9fEhc5YuNMQai2wNZDkwWkQkiko43KL7YV2cxMNtNXw68rqrqyq90Z2lNACYDy6JsM4SqTlDVYlUtxhtn+Z4/eZjD09TSxv/+dQPXXTKVN0t3s3d/EwBfuWRqnHtmjElEXSYQN6ZxHd6ZT+uBJ1V1rYjcIiKXumoLgAIRKQOuB+a6ZdcCTwLrgJeBa1W1LVKbACLyfRGpwNsrWSUiD/be5prOvLJuF1v3NgDw53W7+N1fN9La1h7nXhljElU0h7BQ1SXAEl/ZzUHTB4ArIix7K3BrNG268ruBu7voz1XR9Lu/Wrh06yFls04fd1htrtxSxdKNeznjiAI2A7PPLOaRdzezdse+w2rXGNN/RZVATP/W1NrGjU+vJjcrjX84Zji7Zsxk8vAchg5KY/nmKspmzIx3F40xCchuZWJ4fOlWynbvZ8ZJo8hIC/DtebeTIsKp4/PZWFnPS//6X/HuojEmAVkCGeBUlceXlXPi2DyOGjEEgJtmXwTAqeOHIsAlcy6JYw+NMYnKDmENQMFjKOVVDZTuquMrJ312Hefm0jUA5GalcdSIwRRvKeVv7UogJdzlO8aYgcr2QAa45ZurSAsIx4/JDTu/ZHw+ABv37O/LbhljkoAlkAGsqbWNVdtqOWF0HplpgYPleYXDDk5PLMpmV04+W6sa4tFFY0wCs0NYA9jqilqaW9spKR4aUn7vi8sPTmemBbhk7hMMrWrs6+4ZYxKc7YEMYCu2VFOUk8G4/EEh5U8/MD8kvuGdxymvbsC7uYAxxngsgQxQu/YdYGtVAyXFQ3H3FzvomQV3hcRfX/IQDc1tVDe09GUXjTEJzhLIALVySzUpAiePG9p1Zae82sZBjDGfsQQyALW2t/P+1mqOGTmEnIzohsFSU4QKG0g3xgSxBDIArd9RR0NzG6cV54ed/4uHXwyJX3r4RUbnZVFebQPpxpjPWAIZgFZuqSI3K41Jw3KiXmbM0Cy21zTS1m4D6cYYjyWQAaaiuoFPd+3n1PFDSZHwV5b/9KqLQ+LpV13MmPxBtLYrO/cd6ItuGmOSgCWQAeaRdzYjAiXjox88Bxg71DvVt8IG0o0xjiWQAaTuQAuLlpVz3Ohc8gald2vZoYPSGJQeoMIuKDTGOJZABpAnlpdT19TK5ycVdlrvq3N+GBKvnvNDRISRuZnsqrNDWMYYT1QJRESmiUipiJSJyNww8zNE5Ak3f6mIFAfNm+fKS0Xkwq7aFJHrXJmKSGFQ+TdEZJV7vSMiJ/Z0owei1rZ2fv/2ZqZOyGfM0EGd1v3ad34UEq92cdHgTCrrmuyKdGMMEEUCEZEAcC8wHZgCzBSRKb5qc4BqVZ0EzAfucMtOAa4EjgWmAfeJSKCLNt8GLgC2+NaxCThbVU8Afg7c381tHdCWrNnJtppGvvOFiV3Wvfbi00Liy1w8bHAGTa3tNpBujAGi2wOZCpSp6kZVbQYWATN8dWYAj7jpp4Dzxbs/xgxgkao2qeomoMy1F7FNVf1AVTf7O6Gq76hqtQvfA8Z0YzsHtAMtbfz6lVImDcvh/KOHdVm/Zs/ukHiQi4sGZwBQtttu7W6MiS6BjAbKg+IKVxa2jqq2ArVAQSfLRtNmZ+YAL4WbISLXiMgKEVlRWVnZjSb7r3vfKGNrVQO3zDiWlMN4KNQwl0A2WAIxxhBdAgn3ieM/CB6pTnfLu+6MyLl4CeTGcPNV9X5VLVHVkqKiomia7NfKdtfxu7c28NWTR/O5IzofPO9QfNRxIXGVi3MyUslMS6Gs0hKIMSa654FUAGOD4jHA9gh1KkQkFcgFqrpYtqs2DyEiJwAPAtNVdW8UfR/QWtva+fdn1zAoPZV/v+iYqJe79ZE/hcQvu1hEKMrJsENYxhgguj2Q5cBkEZkgIul4g+KLfXUWA7Pd9OXA6+qdqrMYuNKdpTUBmAwsi7LNECIyDngG+JaqfhLd5g1cbe3KDX/8iGWbqviPi6dQmJMR9bIP3hZ6ot3UoHjY4EzKdtf3Wj+NMcmrywTixjSuA14B1gNPqupaEblFRC511RYABSJSBlwPzHXLrgWeBNYBLwPXqmpbpDYBROT7IlKBt1eySkQedOu4GW9c5T4R+VBEVvTC9vdLqspPn1vN8x9u59+mHcXlp3bvfIM3nn88JJ4UFBcNzmDP/iZq7dkgxgx4Ud3LW1WXAEt8ZTcHTR8Aroiw7K3ArdG06crvBu4OU/5t4NvR9Hcga2ptY94zq3nm/W1cd+4kvnfOpF5tv2MgvayyjlPHh7+brzFmYLAr0fuRqvpmvvXgMp55fxs/uuBIbviHI3t9HXYqrzGmQ3RPEzIJ70BLG7MeeI+Ne+q5e+bJXHriqB63dc8Ly0LiZ4PiodnppKemsKHSxkGMGehsD6SfuOPlj/l4Zx2/++Yph5U8ADZ9vDokzg+KU0SYWJhteyDGGEsg/cEnu+r4/dubuepzxZx39PDDbu/On8wJic/2xZOG5VgCMcZYAkl2jc1tPL2ygiOH5zB3+tF9ss5Jw3Ior27gQEtbn6zPGJOYLIEkuTXbaqlrauX2r51AZlqgT9Y5sSgHVdiy1x4uZcxAZgkkyX1UUUNhTjonj83rtTbnzL0tJF7qiycWZgOwaY8dxjJmILMEksT2HWhh0556ThiTh0R4vnlPnHfZrJB4gy8uPphAbA/EmIHMEkgSW7OtFgWOH53bq+1+44zxIfEsX5yTkUrR4AzbAzFmgLMEksRWVdQyYkgmw4dk9vm6JxRms2mPXQtizEBmFxImqer6ZrZWNfAPU7zTdhcu3Rq23qzTx8Vk/RMLs3l1/a6YtG2MSQ62B5KkVm+rBeCEMb03eN7h5LPOD4m3+WLw9kD27G9m3wG7qaIxA5UlkCS1eW89RYMzyM9O7/W2f3znQyHxW74YPhtI32yHsYwZsCyBJKltNY2MycuKSdu/vuHqkPhsXwzBp/JaAjFmoLIEkoT2HWih7kAro2KUQD54+7WQeLQvBhhXMAgR2Gg3VTRmwLIEkoS21zQCxCyBRCMjNcDovCw277UEYsxAZQkkCW2vaUSAUbl9f/puMDuV15iBLaoEIiLTRKRURMpEZG6Y+Rki8oSbv1REioPmzXPlpSJyYVdtish1rkxFpDCoXETkbjdvlYic0tONTnbbag5QkJNBRozuffXYe1tC4oW+uMPEwmw2VdajqjHphzEmsXWZQEQkANwLTAemADNFZIqv2hygWlUnAfOBO9yyU4ArgWOBaXjPMw900ebbwAWA/1NrOjDZva4Bftu9Te0/ttc0Miovdnsfrz+3MCQ+whd3mFCYTV1TK3v2N8esL8aYxBXNHshUoExVN6pqM7AImOGrMwN4xE0/BZwv3s2ZZgCLVLVJVTcBZa69iG2q6gequjlMP2YAf1DPe0CeiIzszsb2B3v3N1Hb2MLoGI5/LLh9Xkh8ui/ucPBUXhsHMWZAiiaBjAbKg+IKVxa2jqq2ArVAQSfLRtNmT/qBiFwjIitEZEVlZWUXTSafNdv3AfEdQO8wsTAHgE12JpYxA1I0CSTcbV79B70j1elu+eH2A1W9X1VLVLWkqKioiyaTzxp3Bfqo3PgnkNFDs0gLCBttIN2YASmaBFIBjA2KxwDbI9URkVQgF6jqZNlo2uxJP/q9Ndtqyc9OJys9dg+PuuFXC0Lit3xxh0CKMC5/kN2V15gBKpoEshyYLCITRCQdb1B8sa/OYmC2m74ceF29U3MWA1e6s7Qm4A2AL4uyTb/FwD+5s7HOAGpVdUcU/e9XVm+rjen4B8CEo48Piat8cUjdwhw223NBjBmQukwgbkzjOuAVYD3wpKquFZFbRORSV20BUCAiZcD1wFy37FrgSWAd8DJwraq2RWoTQES+LyIVeHsYq0TkQbeOJcBGvIH4B4DvHfbWJ5mG5lYqqhsZEePrP667ZGpI/BVfHGxiUTab9tbT3m6n8hoz0ER1O3dVXYL3AR5cdnPQ9AHgigjL3grcGk2brvxu4O4w5QpcG01/+6uOZ5AXxOAGij01oTCb5tZ2ttc2MmbooHh3xxjTh+xK9CTyWQLJiHNPPlNcYDdVNGagsgSSRLa46y1icQv3YOfOmBkSl/niYBOL7LbuxgxUlkCSyJaqBoYOSovpGVgA3553e0i8zBcHGzY4g0HpATuV15gByBJIEtmyt55x7pBRLN00+6KQeJovDiYiFBfYTRWNGYjsmegJLvhZ5+u272NcfuwHqjeXrgmJ832x34Si7IMXOBpjBg7bA0kSre3t1DS0kJ9AA+gdJhZmU1HdSHNre7y7YozpQ7YHkiRq6ltQoCCnewPowXsw0corHBYSN/hivwmF2bS1K+XVDRxRlNPt9RljkpPtgSSJvfVNQN9cA3Lvi8tD4ud8sV/HXXntporGDCyWQJLE3nrvmRuxPoUX4OkH5ofEx/tiv4mFdi2IMQORJZAksbe+mfRACjkZsT/q+MyCu0Li432xX96gdIYOSmOTPRfEmAHFxkCSRNX+Zgpy0vGe0xVf4cZVBmem2SEsYwYY2wNJEnvrm/vk8FVPFeZkUFZpt3U3ZiCxBJIE2lWpbmjus5so/uLhF0Pil3xxOCOGZFBZ18Se/U2x6pYxJsFYAkkCtY0ttLVrQl4D0mGEe0Lixzvq4twTY0xfsQSSBKrcGVjdvQakp3561cUh8XRfHM5I94yS9Tv2xaRPxpjEYwkkCXQkkPxBiTsGkp2RyvAhGZZAjBlALIEkgZqGFgQYkpUW76506piRQ1hnCcSYASOqBCIi00SkVETKRGRumPkZIvKEm79URIqD5s1z5aUicmFXbbrnpC8VkU9dm+mufJyIvCEiH4jIKhH58uFseDKpbWxmSFYagZS+OYX3q3N+GBKv9sWRHDNyCBsq99s9sYwZILpMICISAO4FpgNTgJkiMsVXbQ5QraqTgPnAHW7ZKcCVwLHANOA+EQl00eYdwHxVnQxUu7YBfor37PSTXZv39WyTk09NQwu5fbj38bXv/CgkXu2LIzlm5BBa2pSy3XY6rzEDQTR7IFOBMlXdqKrNwCJghq/ODOARN/0UcL54V7zNABapapOqbgLKXHth23TLnOfawLV5mZtWYIibzgW2d29Tk1dNYwt5g/ougVx78Wkh8WW+OJJjRgwGbCDdmIEimgQyGigPiitcWdg6qtoK1AIFnSwbqbwAqHFt+Nf1M+CbIlIBLAH+NVxnReQaEVkhIisqKyuj2LzE1q7Kvsa+3QOp2bM7JB7kiyOZUJhNemoKH++0BGLMQBBNAgl34F2jrNNb5QAzgYdVdQzwZeBRETmk/6p6v6qWqGpJUVFRmOaSS31TK63tSl6CD6ADpAZSOGr4YNbbtSDGDAjRJJAKYGxQPIZDDx8drCMiqXiHmKo6WTZS+R4gz7XhX9cc4EkAVX0XyAQKo+h/UqttbAG8Gxb2leKjjguJq3xxZ44ZOZj1O/ah6v+OYYzpb6JJIMuBye7sqHS8AezFvjqLgdlu+nLgdfU+QRYDV7qztCYAk4Flkdp0y7zh2sC1+byb3gqcDyAix+AlkOQ/RtWFmgYvgfTlIaxbH/lTSPyyL+7MMSOHsLe+mco6u6WJMf1dlwnEjUdcB7wCrMc7E2qtiNwiIpe6aguAAhEpA64H5rpl1+LtNawDXgauVdW2SG26tm4ErndtFbi2AW4AviMiHwGPA1fpAPiae3APpA8TyIO3hZ6pPfW2Q87cjujYUbkArKqwZ6Qb099FdTt3VV2CN3AdXHZz0PQB4IoIy94K3BpNm658I95ZWv7ydcBZ0fS3P6lpaCYtIGSlB/psnW88/zjfnnf7wXjS84+zLCjuzAljcklPTeG9jXu5YMrwWHXRGJMA7Er0BFfb2EJeVmI8ByQamWkBThmXx7sb98a7K8aYGLMEkuBqGlvI7cNrQHrDmRMLWbdjHzUNzfHuijEmhiyBJLjahpY+P4X3nheWhcTP+uKunHlEAaqwdFNVb3bLGJNgLIEksKbWNuqaWvt8D2TTx6tD4nxf3JUTx+aSmeaNgxhj+i9LIAlsV613Kmxf74Hc+ZM5IfHZvrgrGakBSsbn8+4GSyDG9GeWQBLYtppGAHKzEvc5IJGceUQBH++sO/gsE2NM/2MJJIFtdwmkL2+k2FvOmFgAwFI7jGVMv2UJJIHtqO3YA+nbBDJn7m0h8VJfHI0TxuQyKD3A2xv29Fa3jDEJxhJIAttWc4Ds9ABpgb79NZ132ayQeIMvjkZaIIWzjyzi5TU7aWmzB0wZ0x9FdSW6iY/tNY19ehPFDt84YzyPvbflYDzrjPEsDIojWbh0a0hcmJPBnv3N/O3TSs472q5KN6a/sT2QBLajtrHPD1/1psnDc8jPTufp97fFuyvGmBiwBJKgVJVt1Y1JdxV6sNSUFC49cRR/Wbfr4E0hjTH9hyWQBLXvQCv1zW1xeZDUyWedHxJv88Xd8dVTRtPc2s6S1TsOt1vGmARjCSRBfXYKb9+Pgfz4zodC4rd8cXccPzqXScNyeOb9isPtljEmwVgCSVDxOoUX4Nc3XB0Sn+2Lu+PxZeUcUZjN8s3VzP/LJyxcuvWQwXZjTHKyBJKgttUcAPr+NiYAH7z9Wkg82hd316nF+aSmCO/YrU33f2/GAAAfiElEQVSM6VeiSiAiMk1ESkWkTEQOeTyde2TtE27+UhEpDpo3z5WXisiFXbXpHnO7VEQ+dW2mB837RxFZJyJrRWRhTzc6GWyvaSQtIORkJv+Z1jkZqZw0No8PtlZT39Qa7+4YY3pJlwlERALAvcB0YAowU0Sm+KrNAapVdRIwH7jDLTsF73nnxwLTgPtEJNBFm3cA81V1MlDt2kZEJgPzgLNU9Vjghz3e6iSwo6aREbmZpCTJg6S6ctakQlrblWWb7RbvxvQX0eyBTAXKVHWjqjYDi4AZvjozgEfc9FPA+eI9Qm8GsEhVm1R1E1Dm2gvbplvmPNcGrs3L3PR3gHtVtRpAVXd3f3OTx/aaA4zMzYrLuh/zXTQYzUWEXRk+JJPJw3J4b8NeWu3KdGP6hWgSyGigPCiucGVh66hqK1ALFHSybKTyAqDGteFf15HAkSLytoi8JyLTouh70tpW08jovPgkkNefCz06eMRzvXO08KxJhdQ1tbJ6W22vtGeMia9oEki4YygaZZ3eKgfvtiuTgXOAmcCDIpJ3SGdFrhGRFSKyorKyMkxzia+tXdm17wCj8jLjsv4Ft88LiU/3xT01eVgOhTnp9qRCY/qJaBJIBTA2KB4DbI9UR0RSgVygqpNlI5XvAfJcG/51VQDPq2qLOxxWipdQQqjq/apaoqolRUVFUWxe4qmsa6K1XeN2CCtWRISpxflsrWrg45374t0dY8xhiiaBLAcmu7Oj0vEGxRf76iwGZrvpy4HXVVVd+ZXuLK0JeB/4yyK16ZZ5w7WBa/N5N/0ccC6AiBTiHdLa2N0NTgYdD5KK1yGsWDpl3FBSU8SuBTGmH+gygbjxiOuAV4D1wJOqulZEbhGRS121BUCBiJQB1wNz3bJrgSeBdcDLwLWq2hapTdfWjcD1rq0C1zau7l4RWYeXZH6iqv3ywoKOq9BHxSmB3PCrBSHxW774cAzKSOW40bk8+/42GprtlF5jkllUFxmo6hJgia/s5qDpA8AVEZa9Fbg1mjZd+Ua8s7T85YqXnK6Pps/JrOMq9FF5maw8/BOgum3C0ceHxFW++HBNLc7nw/IaXvhoO18/bVyvtm2M6Tt2JXoC2l5zgMEZqQzOjM+deK+7JDR/f+WSQ/L5YRlfMIjJw3JYuKy868rGmIRlCSQBbatpjNvhq74gIlxRMoaPymvYvKc+3t0xxvSQJZAEtKO2MW6n8PaVi04YBcCLq/wn9BljkoUlkAS0veYAI+O4B3LujJkhcZkv7g2j87IoGT+UFz6y54QYk6wsgSSYxuY2quqb43oK77fn3R4SL/PFveXSk0ZRuquO0p11MWnfGBNblkASzHZ3BtbI3Pgdwrpp9kUh8TRf3FumHzeSFIEXPrLDWMYkI0sgCaai2ksgY/MHxa0Pm0vXhMT5vri3FA3O4HNHFPLCqu14Z2kbY5KJJZAEs7WqAYBxcUwgfenSE0exZW+D3WDRmCRkCSTBlFc1kJ6aQlFORtz6kFc4LCRu8MW96cJjR5AWEDuMZUwSsgSSYLbubWDs0CxSUuL3IKl7X1weEj/ni3tT7qA0zj6yiBdX7aC93Q5jGZNMLIEkmPLqhrgfvnr6gfkh8fG+uLddcuIodtQeYMWW6piuxxjTuyyBJBBV9fZA4pxAnllwV0h8vC/ubRccM5zMtBQ7jGVMkrEEkkBqG1uoa2qN+x5IX8vOSOX8Y4azZPUOe9ytMUnEEkgCKa+K/ym88XLJCaPYW9/MOxv65R36jemXLIEkkI5TeMcOjW8C+cXDL4bEL/niWDjnqCIGZ6Ty3IfbYr4uY0zviOp5IKZvHEwg+f33TryRZKYFOHrkEF74aDvHjcolMy0AwKzT7XkhxiQq2wNJIOXVDeRnp8ftOSAdfnrVxSHxdF8cK6cVD6WlTfmwvKZP1meMOTxRJRARmSYipSJSJiJzw8zPEJEn3PylIlIcNG+eKy8VkQu7atM9J32piHzq2kz3retyEVERKenJBiey8qr4n4HVVxYu3XrIa3ReFiNzM1mxuSre3TPGRKHLBCIiAeBeYDowBZgpIlN81eYA1ao6CZgP3OGWnQJcCRwLTAPuE5FAF23eAcxX1clAtWu7oy+Dge8DS3u2uYlta5V3EeFAJSKcVpzP9toDbHP3BDPGJK5o9kCmAmWqulFVm4FFwAxfnRnAI276KeB8ERFXvkhVm1R1E1Dm2gvbplvmPNcGrs3Lgtbzc+CXwIFubmfCa2tXtlU3JsQpvF+d88OQeLUvjqWTxuaRFhCW216IMQkvmgQyGgh+eHWFKwtbR1VbgVqgoJNlI5UXADWujZB1icjJwFhV7fSUIBG5RkRWiMiKysrKKDYvMeyobaS1XRMigXztOz8KiVf74ljKTAtw/OhcPqqoobG5rc/Wa4zpvmgSSLibMvlvWhSpTq+Ui0gK3qGxGzrpp1dZ9X5VLVHVkqKioq6qJ4zPzsCKfwK59uLTQuLLfHGsfX5SEc2t7bz1ye4+Xa8xpnuiSSAVwNigeAzgv+fEwToikgrkAlWdLBupfA+Q59oILh8MHAe8KSKbgTOAxf1pIL08gW7jXrMn9IN70J6+/SAfkZvJSWPzeGfDXrbX2FiIMYkqmgSyHJjszo5KxxsUX+yrsxiY7aYvB15X7wlBi4Er3VlaE4DJwLJIbbpl3nBt4Np8XlVrVbVQVYtVtRh4D7hUVVf0cLsTTnlVI4EUieuTCBPJBVOGo8Bdr34S764YYyLoMoG48YjrgFeA9cCTqrpWRG4RkUtdtQVAgYiUAdcDc92ya4EngXXAy8C1qtoWqU3X1o3A9a6tAtd2v7ehcj/j8geRGoj/pTnFRx0XElf54r4wdFA6Z0zI56mVFfbMdGMSlPTnR4mWlJToihXJsZNy3p1vMnlYDv/7rdCjcguXbo1Tj+KvvqmV+94sY/iQTJ793llkpQfi3SVjBgQRWamqXQ4RxP/rruFASxtb9jZw5PDB8e4KAA/eFnqt6NTbDrl2tE9kZ6Qy/+snUbqrjv94fk3Ic9PDXYhojOlblkASwMbKetraNWESyBvPPx4ST/LFfemco4bxr+dN5qmVFSxaXt71AsaYPmMJJAF8uts7xn/UiMRIIInmB+dP5guTC7n5+TW8U7Yn3t0xxjiWQBJA6c46UlOE4oLseHclIQVShHtmncKEwmy+++hKG1Q3JkFYAkkAn+yqY2JRNumpifHruOeFZSHxs744HnKz0vj9P09lUEaAq36/jLoDLfHukjEDnj0PJAF8sms/uVlpCTMQvOnj1QwtGn4wzv94NduC4ngZnZfFQ1edxtd++w5Prijnn8+aQIqEu3mBMaYvJMZX3gGsobmVrVUNDB+SEe+uHHTnT+aExGf74r7kP9Pqo/JavnzcSDZU1vPWJ8lzrzNj+iNLIHFWtns/AMMG2xXo0Tp1/FBOGJPLq+t2sWlPfby7Y8yAZQkkzjoGhEcMsQQSLRHhspNGMzQ7nWfer6C1rT3eXTJmQLIEEmef7t5PemoK+TnpXVfuI3Pm3hYSL/XFiSAzLcCME0ext76Zv35qp/YaEw+WQOKsdGcdk4pyEmow+LzLZoXEG3xxopg8fDDHjRrCm6W7qapvjnd3jBlwLIHE2ae76hLuAsJvnDE+JJ7lixPJRSeMIkWEF1f5nzBgjIk1SyBxtHvfAbbXHuDYUUPi3ZWklZuVxnlHD+PjnXX81c7KMqZPWQKJo6WbvOd+T52QH+eeJLfPHVFAfnY6t/5pPW3t/ffu0sYkGksgcbRsUxXZ6QGmjEysPZCTzzo/JN7mixNNaiCFaceOoHRXHU+usBsuGtNXLIHE0bJNVZxanJ8QD5EK9uM7HwqJ3/LFiejYUUM4rXgod/65lP1NrfHujjEDQmJ9cg0g1fXNlO6q4/QEPHz16xuuDonP9sWJSET4j4unsGd/M3e/9mm8u2PMgBBVAhGRaSJSKiJlInLI04XcM8+fcPOXikhx0Lx5rrxURC7sqk33nPSlIvKpazPdlV8vIutEZJWIvCYiiXtqUBSWb07c8Y8P3n4tJB7tixPVCWPymDl1LAv+vomPd+6Ld3eM6fe6TCAiEgDuBaYDU4CZIjLFV20OUK2qk4D5wB1u2SnAlcCxwDTgPhEJdNHmHcB8VZ0MVLu2AT4ASlT1BOAp4Jc92+TEsGxTFempKZwwJjfeXelXbpx2NLlZadz07BrabUDdmJiKZg9kKlCmqhtVtRlYBMzw1ZkBPOKmnwLOFxFx5YtUtUlVNwFlrr2wbbplznNt4Nq8DEBV31DVBlf+HjCm+5ubOJZtruLksXlkpNpzvntT3qB0/v3Lx7ByS7UNqBsTY9EkkNFA8H9ihSsLW0dVW4FaoKCTZSOVFwA1ro1I6wJvr+SlcJ0VkWtEZIWIrKisTMzrAvY3tbJmW21Cjn8APPbelpB4oS9OdF87ZTSnT8jnF39af/BmlcaY3hdNAgl3jw3/sYFIdXqr/LMViXwTKAF+FaYuqnq/qpaoaklRUVG4KnG3fHMV7QpTJxTEuythvf7cwpD4CF+c6ESE+V8/iYzUFL776Ao7K8uYGIkmgVQAY4PiMYD/vhEH64hIKpALVHWybKTyPUCea+OQdYnIBcBNwKWq2hRF3xPSy6t3kp0e4NTxQ+PdlbAW3D4vJD7dFyeDUXlZ/M+sk9m8t4EfP/mRjYcYEwPRJJDlwGR3dlQ63qD4Yl+dxcBsN3058Lqqqiu/0p2lNQGYDCyL1KZb5g3XBq7N5wFE5GTgf/GSx+6ebW78HWhpY8nqHUw7biRZ6Tb+EUufO6KQedOP5uW1O/neY+/T0Gx7Isb0pi4TiBuPuA54BVgPPKmqa0XkFhG51FVbABSISBlwPTDXLbsWeBJYB7wMXKuqbZHadG3dCFzv2ipwbYN3yCoH+KOIfCgi/iSWFN74eDd1Ta1cdvKoeHdlQJjz+Qn8x8VT+PO6nXztt++yodLGRIzpLeJ96e+fSkpKdMWKFfHuRohr/rCCD8preG/e+QRSPhvySZTnoQO8/7dXOeULFxyMR//tVbYFxYlq1unjIs57o3Q331/4AXVNrXzxyCJGDMkgPzuDIZmpB+8EkJGawuzPFfdRb41JXCKyUlVLuqqX2lUF03tqGpp5s7SSb505PiR5JJoJRx8fElf54mR07lHDeO2Gs3l8WTmPL9sa8c6997xRxoTCbK44dQyXnjSKjNRA2OTeWbIyZqCwBNKHlqzeSXNbO185OdyZyYnjukumhpzK+5VLpibFqbyR9uI6PuyHDcnkBxdM5tpzj+A3r31KbWMLdY2ttKmieONT+YPSeX9rNT95ahW/fKWUudOO7sMtMCa5WALpI6rKEyvKOaIo257/EWepgRRG5mYxMjfrkHmzTh+HqvL3sj385tVPueGPH/H5SYVMO25EQj010phEYAkkhoK/EX+yq46Pymu47KTRiH0QJazg39mMk0aTGkjh72V72LO/iVmnjyM1xe4/akwHSyB9QFV5bf0u8rLSOGV8XkINmIdz7oyZIXGZL042PX2/AynCpSeOoignnRdW7eCZ97dx+aljbE/EGMcSSB/4dPd+yqsbueyk0UnxDfbb824PiZf54oHmzCMKaWpt58/rdjE4M5Xpx42Md5eMSQiJ/2mW5Px7H8ngptkXhcTTfPFAdPaRRZwxMZ+/fbqHdzfujXd3jEkIlkBibPW2WsqrGznnqGFJsfcBsLl0TUic74sHIhHh4hNGccyIwfxp1XbeinAasDEDSXJ8oiWphuZWXli1g9F5WZQUJ+Z9r0z0UkT4x9PGMmxwJtc99j6f7qqLd5eMiStLIDH00pqdNDa38tVTRifVwGte4bCQuMEXD2QZqQH+6czxZKQFmP3QMsqrGrpeyJh+yhJIjLyzYQ8rt1Tz+UlFYa83SGT3vrg8JH7OFw90eYPSeeTq06hvbmPmA++xraYx3l0yJi4sgcTA7roD/HDRhxRkp3Pe0cn37f3pB+aHxMf7YgPHjsrl0TlTqW1sYeb977F+R+RnsC9cujXsy5hkZwmkl7W0tXPdwg/Yd6CFWaePIz01+d7iZxbcFRIf74uN54Qxefzh6qk0NLdx6T1/57dvbqDNnjtiBhC7DqSX3fHSxyzbVMX8r59IY3N7vLtjYiR4D+K7X5zIcx9u446XP+b3b29i+nEj+OKRRRQNzmDooHT2NbYgAplpAdICyfeFwphILIH0ElXlrlc/5cG/b2L2meP5yslj7DDFAJGdkcqsqeNYv6OOyv0HWLS8nEfeDX/zyez0AEOz0ykuyGZkbiZnTCywB4uZpGUJpBeoKr/+cyn3vrGBK04dw82XHBvvLh2WXzz8Ykj8ki82hxIRpowawqzTj6O+qZWPd+6jur6F6oZm3i7bS5sqjc1t1Da2sGd/E+9u3Mvfy/aQmZbCOUcOY9pxIzj36GHkZqXFe1OMiZolkMO0u+4ANz+3lpfX7mTm1HHcetlxpCTwsz5M7GVnpHLq+PyDcUvboeMiLW3tTCzK5i/rdvHK2p28vHYnaQHhzCMKOWVcHkePGMzovEFkZwTISAvQ2tZOc2s7ze5nS5vy8pqdtLUrIt41KikpMO3YEaSkCKkpQiBFSE1JIZAi/HntTrIzUkMOodkzTczhiiqBiMg04DdAAHhQVW/3zc8A/gCcCuwFvq6qm928ecAcoA34vqq+0lmb7tnpi4B84H3gW6ra3Nk64qG2sYXnPtjG/Fc/oaG5jbnTj+aaL0zsF8njp1ddHPI8kOlXXZwUzwNJJmmBFMqrGjl6xBCOHD6YiupG1m6vZXtNI3/7tJKePij0wb9t6nR+VlqAYYMzGJGbCcAxIwdz1IjBDErvve+S9gCugaPLvxoRCQD3Al8CKoDlIrJYVdcFVZsDVKvqJBG5ErgD+LqITAGuBI4FRgGvisiRbplIbd4BzFfVRSLyO9f2byOt43DfgM6oKk2t7exvaqWmoYUdtY2s37GPD8treHX9bppb2ykZP5Tbv3YCk4blxLIrJkn0ZNwrRYRx+YMYlz8IgObWdnbXHaDuQCtNre20trUTOLhHIW465WAZqrQptKui7md7u3o/Xdzxd1zb2MKufQf4sLyGpZuqABCBCQXZHD1yMBMKsxkzdBAjhmQyJCuVnIw0UgQUaGhuo7qhmdoG79BcdUMLVfVN7N3fzN79zeypb6K6vpmG5jZUISXFS5RpgRT+8O5mMtICZKWlkJkWICstQGZagIrqRtIDQmoghdSAt8c0dUI+6akpZARSyEhLYVB6KllpAbLSU8hKSyUrPeBi72daQBL2EQkdjwxvbmunsbmNhuY2GlvaDk4vWb2DlrZ2WtuU1nZvz7K1XTllXB6ZaQEyUlMO/sxKDzAo3XvfPntP4vseRPO1YypQpqobAURkETADCE4gM4CfuemngHvE25oZwCJVbQI2iUiZa49wbYrIeuA8YJar84hr97eR1qExeKj7ktU7+MGiD8IeegAYPiSDK08by+WnjuH40bkJ+8drklN6agpjhg6K6TpUlS8eWcS6HftY715rt+/jlbW7unUq8tBBaRTkZFCQnc4xI4YwNDuNjZX1pIjQpkpLazstbe2MyM2ksaWdAy1tVNU309jcxoHWNqrrW2hp8+p0rPaVtTu7tS0iIHDw/1BcmTftzZSQunJw2nsv3E/UF3NwItK8jo+fz+JudT2iJat3dHuZlI5DmSKIwHe+MJEfX3hU73QogmgSyGigPCiuAE6PVEdVW0WkFihw5e/5lu14nmu4NguAGlVtDVM/0jr2BHdERK4BrnHhfhEpjWIbu2ULsAz4eddVC/H1L0F02a9vnDH+s2mAoDiGEvX9gsTtW1z71cmBTXu/uqfX+/UT9+qhqP7ho0kg4b5e+/NspDqRysOdDN9Z/Wj7gareD9wfpm6fE5EVqloS7374Wb+6L1H7Zv3qHutX74rmqqYKYGxQPAbYHqmOiKQCuUBVJ8tGKt8D5Lk2/OuKtA5jjDFxEE0CWQ5MFpEJIpKONyi+2FdnMTDbTV8OvO7GJhYDV4pIhju7ajLe0Z+wbbpl3nBt4Np8vot1GGOMiYMuD2G58YbrgFfwTrl9SFXXisgtwApVXQwsAB51g+RVeAkBV+9JvAH3VuBaVW0DCNemW+WNwCIR+QXwgWubSOtIcAlxKC0M61f3JWrfrF/dY/3qRWJf4o0xxvSE3dnNGGNMj1gCMcYY0yOWQGJERKaJSKmIlInI3D5a52YRWS0iH4rICleWLyJ/EZFP3c+hrlxE5G7Xv1UickpQO7Nd/U9FZHak9XXSj4dEZLeIrAkq67V+iMipbjvL3LJRXckZoV8/E5Ft7j37UES+HDRvnltHqYhcGFQe9nfrTgpZ6vr7hDtBJJp+jRWRN0RkvYisFZEfJMJ71km/4vqeiUimiCwTkY9cv/6rs7bEO4nnCbfupSJS3NP+9rBfD4vIpqD36yRX3md/+zGjqvbq5RfeiQEbgIlAOvARMKUP1rsZKPSV/RKY66bnAne46S8DL+FdX3MGsNSV5wMb3c+hbnpoN/vxReAUYE0s+oF3Jt+ZbpmXgOmH0a+fAT8OU3eK+71lABPc7zPQ2e8WeBK40k3/DviXKPs1EjjFTQ8GPnHrj+t71km/4vqeuW3IcdNpwFL3PoRtC/ge8Ds3fSXwRE/728N+PQxcHqZ+n/3tx+pleyCxcfD2L6rajHdzyBlx6ssMvFvC4H5eFlT+B/W8h3f9zUjgQuAvqlqlqtXAX4Bp3Vmhqv6VQ6/R6ZV+uHlDVPVd9f6j/hDUVk/6FcnB2/Co6iag4zY8YX+37pvgeXi32fFvY1f92qGq77vpOmA93p0X4vqeddKvSPrkPXPbvd+Fae6lnbQV/D4+BZzv1t2t/h5GvyLps7/9WLEEEhvhbv/S2T9eb1HgzyKyUrxbugAMV9Ud4H0gAB0PaY/Ux1j1vbf6MdpN92b/rnOHEB7qOEzUg351dhueqLnDKyfjfXtNmPfM1y+I83smIgER+RDYjfcBu6GTtkJugwQE32qpV/8H/P1S1Y7361b3fs0X787iIf2Kcv2x+Ns/LJZAYiOq267EwFmqegowHbhWRL7YSd3u3n4mVrrbj97u32+BI4CTgB3AnfHql4jkAE8DP1TVfZ1V7cu+helX3N8zVW1T1ZPw7lYxFTimk7bi1i8ROQ6YBxwNnIZ3WOrGvu5XrFgCiY1obv/S61R1u/u5G3gW7x9rl9v1xf3c3UUfY9X33upHhZvulf6p6i73T98OPMBnd4vuzdvwdElE0vA+pB9T1Wdccdzfs3D9SpT3zPWlBngTbwyhu7dBitn/QFC/prlDgareXcl/T8/fr1792+8VsRxgGagvvCv8N+INzHUMwh0b43VmA4ODpt/BG7v4FaEDsb900xcROoC3zJXnA5vwBu+Guun8HvSnmNDB6l7rB96tcM7gs4HELx9Gv0YGTf8I75g4eM+wCR5g3Yg3uBrxdwv8kdBB3O9F2SfBO559l688ru9ZJ/2K63sGFAF5bjoL+BtwcaS2gGsJHUR/sqf97WG/Rga9n3cBt8fjbz8mnzvxXHl/fuGdYfEJ3rHZm/pgfRPdH/pHwNqOdeId630N+NT97PhDFLyHem0AVgMlQW1djTegWAb8cw/68jjeoY0WvG9Nc3qzH0AJsMYtcw/ujgo97Nejbr2r8O63FvzheJNbRylBZ7tE+t2638Ey198/AhlR9uvzeIciVgEfuteX4/2eddKvuL5nwAl4tzla5bbp5s7aAjJdXObmT+xpf3vYr9fd+7UG+D8+O1Orz/72Y/WyW5kYY4zpERsDMcYY0yOWQIwxxvSIJRBjjDE9YgnEGGNMj1gCMcYY0yOWQMyAJyIqIo8GxakiUikiL8Z4vWe4u8N+6O54+7Mu6p8T6z4Z0x1dPtLWmAGgHjhORLJUtRH4ErCtD9b7CPCPqvqRiASAo3qzcRFJ1c/uDWVMr7M9EGM8L+FdGQwwE++iQwBEJNvdNHC5iHwgIjNcebGI/E1E3nevz7nyc0TkTRF5SkQ+FpHHIjy3YRjehY2od2uQdW75qSLyjlvXOyJySGKJVEdErhKRP4rIC3g31ny0o79u/mMicmlvvGHGWAIxxrMIuFJEMvGuKF4aNO8m4HVVPQ04F/iViGTj3ZvqS+rdwPLrwN1By5wM/BDvmRMTgbPCrHM+UCoiz4rId926AT4GvqiqJwM3A/8vzLKd1TkTmK2q5wEPAv8MICK5wOeAJdG8IcZ0xQ5hGQOo6ip3y/KZHPoB+w/ApSLyYxdnAuPwbmR3j3vCXBtwZNAyy1S1AsDd3rsY+LtvnbeIyGOu/Vlu3efg3ezvERGZjHcrkbQwXe6szl9Utcqt4y0RuVdEhgFfBZ62w1qmt1gCMeYzi4Ff432IFwSVC/A1VS0NruwGvXcBJ+LtzR8Imt0UNN1GhP81Vd0A/FZEHgAqRaQA+Dnwhqp+xSW1N8Ms2lmdel/dR4Fv4N1I8Opw/TCmJ+wQljGfeQi4RVVX+8pfAf61YxxDRE525bnADvVua/4tvDu5Rk1ELgoaG5mMl2hqXLsdg/hXRVg8mjodHsY7nIaqru1OH43pjCUQYxxVrVDV34SZ9XO8Q0SrRGSNiwHuA2aLyHt4h6/83/y78i28MZAPcXsJqtqG9yz020TkbSInpWjqdGzXLrzH0f6+m/0zplN2N15j+jkRGYR3u/BTVLU23v0x/YftgRjTj4nIBXhnbP2PJQ/T22wPxBhjTI/YHogxxpgesQRijDGmRyyBGGOM6RFLIMYYY3rEEogxxpge+f8BZJ0HQGDiz7wAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plotting the distribution of Mean Salary after processing\n",
    "sns.distplot(df[\"Mean Salary\"])\n",
    "plt.title(\"Frequency of Mean Salaries\")\n",
    "plt.axvline(df[\"Mean Salary\"].mean(), color='r', linestyle='dashed', linewidth=1)\n",
    "plt.axvline(df[\"Mean Salary\"].median(), color='black', linestyle='dashed', linewidth=1)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### For the distribution of  Mean Salaries, the mean (red line) is greater than the median (black line), due to the presence of some extreme values.\n",
    "- In this case, I will classify Mean Salary into 2 classes.\n",
    "- Class 0 represents below or equal Median, Class 1 represents above Median.\n",
    "- Reason is to identify if model can predict the correct class given certain features."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Median: 6000.0, Mean: 6582.234043834296\n"
     ]
    }
   ],
   "source": [
    "print(\"Median:\", df[\"Mean Salary\"].median(), end=\", \")\n",
    "print(\"Mean:\", df[\"Mean Salary\"].mean())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "df[\"Salary_Category\"] = 0\n",
    "# Code for 2 categories (Above Median = 1 / Below & Equals Median = 0)\n",
    "df.loc[df[df[\"Mean Salary\"] > df[\"Mean Salary\"].median()].index, \"Salary_Category\"] = 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot to show the distribution of Salary Classes\n",
    "df[\"Salary_Category\"].value_counts().plot.bar()\n",
    "plt.title(\"Salaries Category Distribution\")\n",
    "plt.xlabel(\"Salary Categories\")\n",
    "plt.ylabel(\"Counts\")\n",
    "plt.xticks(rotation=0)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0] 0.5173410404624278, [1] 0.48265895953757226\n"
     ]
    }
   ],
   "source": [
    "print(\"[0]\", len(df[df[\"Salary_Category\"] == 0]) / len(df[\"Salary_Category\"]), end=\", \")\n",
    "print(\"[1]\", len(df[df[\"Salary_Category\"] == 1]) / len(df[\"Salary_Category\"]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Min Salary</th>\n",
       "      <th>Max Salary</th>\n",
       "      <th>Contract</th>\n",
       "      <th>Internship</th>\n",
       "      <th>Temporary</th>\n",
       "      <th>Permanent</th>\n",
       "      <th>Mean Salary</th>\n",
       "      <th>Salary_Category</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>1384.000000</td>\n",
       "      <td>1384.000000</td>\n",
       "      <td>1384.000000</td>\n",
       "      <td>1384.000000</td>\n",
       "      <td>1384.000000</td>\n",
       "      <td>1384.000000</td>\n",
       "      <td>1384.000000</td>\n",
       "      <td>1384.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>6983.350434</td>\n",
       "      <td>10923.443642</td>\n",
       "      <td>0.182803</td>\n",
       "      <td>0.008671</td>\n",
       "      <td>0.009393</td>\n",
       "      <td>0.440029</td>\n",
       "      <td>6582.234044</td>\n",
       "      <td>0.482659</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>13540.037234</td>\n",
       "      <td>21374.532350</td>\n",
       "      <td>0.386645</td>\n",
       "      <td>0.092745</td>\n",
       "      <td>0.096496</td>\n",
       "      <td>0.496570</td>\n",
       "      <td>3613.102381</td>\n",
       "      <td>0.499880</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>7.000000</td>\n",
       "      <td>9.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>8.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>3000.000000</td>\n",
       "      <td>5000.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>4000.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>5000.000000</td>\n",
       "      <td>7500.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>6000.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>6500.000000</td>\n",
       "      <td>10000.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>8000.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>164500.000000</td>\n",
       "      <td>277000.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>35500.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          Min Salary     Max Salary     Contract   Internship    Temporary  \\\n",
       "count    1384.000000    1384.000000  1384.000000  1384.000000  1384.000000   \n",
       "mean     6983.350434   10923.443642     0.182803     0.008671     0.009393   \n",
       "std     13540.037234   21374.532350     0.386645     0.092745     0.096496   \n",
       "min         7.000000       9.000000     0.000000     0.000000     0.000000   \n",
       "25%      3000.000000    5000.000000     0.000000     0.000000     0.000000   \n",
       "50%      5000.000000    7500.000000     0.000000     0.000000     0.000000   \n",
       "75%      6500.000000   10000.000000     0.000000     0.000000     0.000000   \n",
       "max    164500.000000  277000.000000     1.000000     1.000000     1.000000   \n",
       "\n",
       "         Permanent   Mean Salary  Salary_Category  \n",
       "count  1384.000000   1384.000000      1384.000000  \n",
       "mean      0.440029   6582.234044         0.482659  \n",
       "std       0.496570   3613.102381         0.499880  \n",
       "min       0.000000      8.000000         0.000000  \n",
       "25%       0.000000   4000.000000         0.000000  \n",
       "50%       0.000000   6000.000000         0.000000  \n",
       "75%       1.000000   8000.000000         1.000000  \n",
       "max       1.000000  35500.000000         1.000000  "
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Dummy Coding Job Levels\n",
    "list1 = ['Executive', 'Fresh/entry level', 'Junior Executive', 'Manager', 'Middle Management', 'Non-executive', 'Professional', 'Senior Executive', 'Senior Management']\n",
    "for count, x in enumerate(df[\"Job Level\"].apply(lambda x: x.split(\", \"))):\n",
    "    for y in list1:\n",
    "        if y in x:\n",
    "            df.loc[count, y] = int(1)\n",
    "        else:\n",
    "            df.loc[count, y] = int(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Job Title</th>\n",
       "      <th>Job Type</th>\n",
       "      <th>Company Name</th>\n",
       "      <th>Job Location</th>\n",
       "      <th>Job Industry</th>\n",
       "      <th>Contract</th>\n",
       "      <th>Internship</th>\n",
       "      <th>Temporary</th>\n",
       "      <th>Permanent</th>\n",
       "      <th>Job Description</th>\n",
       "      <th>...</th>\n",
       "      <th>Salary_Category</th>\n",
       "      <th>Executive</th>\n",
       "      <th>Fresh/entry level</th>\n",
       "      <th>Junior Executive</th>\n",
       "      <th>Manager</th>\n",
       "      <th>Middle Management</th>\n",
       "      <th>Non-executive</th>\n",
       "      <th>Professional</th>\n",
       "      <th>Senior Executive</th>\n",
       "      <th>Senior Management</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Data Analyst</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>INTERNATIONAL SOS TECHNOLOGY SERVICES PTE. LTD.</td>\n",
       "      <td>ODEON TOWERS, 331 NORTH BRIDGE ROAD 188720</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>in our digital innovation services centre (dis...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Data Analyst</td>\n",
       "      <td>Permanent</td>\n",
       "      <td>ASCENTIS PTE. LTD.</td>\n",
       "      <td>3015A UBI ROAD 1 408705</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>data analyst job description: the role will be...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Datawarehouse Consultant</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>R SYSTEMS (SINGAPORE) PTE LIMITED</td>\n",
       "      <td>HOI HUP BUILDING, 16 JALAN KILANG 159416</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>experience working with business users to con...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Data analyst</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>TRAKOMATIC PTE. LTD.</td>\n",
       "      <td>ALICE@MEDIAPOLIS, 29 MEDIA CIRCLE 138565</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>the retail data analyst will lead and perform ...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Data Analyst</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>GUMI ASIA PTE. LTD.</td>\n",
       "      <td>CONNEXIS, 1 FUSIONOPOLIS WAY 138632</td>\n",
       "      <td>Others</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>perform evaluations and problem investigation...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Senior Associate, Data and Analytics, Advisory</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>ERNST &amp; YOUNG ADVISORY PTE. LTD.</td>\n",
       "      <td>1 RAFFLES QUAY 048583</td>\n",
       "      <td>Consulting</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>powered by big data and advanced technologies,...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>Associate, Advisory Data and Analytics</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>ERNST &amp; YOUNG ADVISORY PTE. LTD.</td>\n",
       "      <td>1 RAFFLES QUAY 048583</td>\n",
       "      <td>Consulting</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>powered by big data and advanced technologies,...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>Data Analyst</td>\n",
       "      <td>Contract</td>\n",
       "      <td>FUTUREX TECHNOLOGIES PTE. LTD.</td>\n",
       "      <td>GERMAN CENTRE, 25 INTERNATIONAL BUSINESS PARK ...</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>the successful candidate will be responsible f...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>Senior Data Scientist</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>SINGAPORE POWER LIMITED</td>\n",
       "      <td>SP GROUP BUILDING, 2 KALLANG SECTOR 349277</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>why work for us we power the nation.  make the...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>Data Analyst, Transformation Project</td>\n",
       "      <td>Contract</td>\n",
       "      <td>CHANDLER MACLEOD GROUP PTE. LTD.</td>\n",
       "      <td>KECK SENG TOWER, 133 CECIL STREET 069535</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>23 months contract global bank, based in cbd ...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>Commercial Data Analyst (East, Advanced Excel,...</td>\n",
       "      <td>Permanent, Full Time</td>\n",
       "      <td>GMP RECRUITMENT SERVICES (S) PTE LTD</td>\n",
       "      <td>ONE FINLAYSON GREEN, 1 FINLAYSON GREEN 049246</td>\n",
       "      <td>Banking and Finance</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>a well-established fmcg.  responsibilities:​​...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>Senior Data Analyst</td>\n",
       "      <td>Permanent, Full Time</td>\n",
       "      <td>99 PTE. LTD.</td>\n",
       "      <td>71 AYER RAJAH CRESCENT 139951</td>\n",
       "      <td>Engineering</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>99.co is looking for a senior data analyst to ...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>Data Analyst (SAS, R or Python)</td>\n",
       "      <td>Contract</td>\n",
       "      <td>OPTIMUM SOLUTIONS (SINGAPORE) PTE LTD</td>\n",
       "      <td>PLAZA 8 @ CBP, 1 CHANGI BUSINESS PARK CRESCENT...</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>optimum solutions (company registration number...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>Data Analyst</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>MINDTREE LIMITED (SINGAPORE BRANCH)</td>\n",
       "      <td>HONEYWELL BUILDING, 17 CHANGI BUSINESS PARK CE...</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1) around 6+ years of it experience across var...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>Data Analyst</td>\n",
       "      <td>Contract</td>\n",
       "      <td>HYDROGEN CONSULTING SOLUTIONS PTE. LTD.</td>\n",
       "      <td>PRUDENTIAL TOWER, 30 CECIL STREET 049712</td>\n",
       "      <td>Banking and Finance</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>min. 2 years’ experience in banking industry p...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>Business Data Analyst</td>\n",
       "      <td>Permanent, Contract</td>\n",
       "      <td>RGF TALENT SOLUTIONS SINGAPORE PTE. LTD.</td>\n",
       "      <td>71 ROBINSON ROAD 068895</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>primarily support the global marketing team w...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>Data Analyst</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>Company Undisclosed</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Others</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>develop and implement data collection, data a...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>EBS Spark Developer</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>CITTABASE SOLUTIONS PTE. LTD.</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>oracle ebs data warehousing reporting and anal...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>Wealth Management – Senior Data Analyst, Clien...</td>\n",
       "      <td>Permanent, Full Time</td>\n",
       "      <td>DEUTSCHE BANK AKTIENGESELLSCHAFT</td>\n",
       "      <td>1 RAFFLES QUAY 048583</td>\n",
       "      <td>Banking and Finance</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>private &amp; commercial bank deutsche bank's newl...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>DATA  ANALYST</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>Company Undisclosed</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>identify valuable data sources and automate c...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>Business Analyst Project Manager - Data Science</td>\n",
       "      <td>Permanent, Contract</td>\n",
       "      <td>Company Undisclosed</td>\n",
       "      <td>HONG LEONG BUILDING, 16 RAFFLES QUAY 048581</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>who we are:  we ns solutions asia pacific (nsa...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>Global CDS Operations OneTouch Data Analyst</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>BCD TRAVEL ASIA PACIFIC PTE. LTD.</td>\n",
       "      <td>TWENTY ANSON, 20 ANSON ROAD 079912</td>\n",
       "      <td>Travel / Tourism</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>where willyourcareer take you?     we’re not j...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>Data Management Analyst</td>\n",
       "      <td>Contract, Full Time</td>\n",
       "      <td>MANPOWER STAFFING SERVICES (SINGAPORE) PTE LTD</td>\n",
       "      <td>GUOCO TOWER, 1 WALLICH STREET 078881</td>\n",
       "      <td>Advertising / Media , Customer Service</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>we are looking for new members to help us achi...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>Data Analyst</td>\n",
       "      <td>Permanent</td>\n",
       "      <td>ABAKUS (ASIA PACIFIC) PTE. LTD.</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>to reinvent an industry, you need to build an ...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>Senior / Data Analyst (1 Year Contract) - S$3000</td>\n",
       "      <td>Contract</td>\n",
       "      <td>CLARIST RESOURCES PTE. LTD.</td>\n",
       "      <td>ORCHARD TOWERS, 400 ORCHARD ROAD 238875</td>\n",
       "      <td>Information Technology, Logistics / Supply Cha...</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>our client for this role well-established, hom...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>Data Analyst (Stack) - 1 year contract, Group ...</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>OVERSEA-CHINESE BANKING CORPORATION LIMITED</td>\n",
       "      <td>63 CHULIA STREET 049514</td>\n",
       "      <td>Others</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>job description:   work alongside the group cu...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>Senior Analyst, Data Science</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>LAZADA SOUTH EAST ASIA PTE. LTD.</td>\n",
       "      <td>AXA TOWER, 8 SHENTON WAY 068811</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>team introduction lazada is the number one onl...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>Informatica Developer</td>\n",
       "      <td>Permanent</td>\n",
       "      <td>ADDSTONES SAS</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Consulting , Banking and Finance, Information ...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>gfi group is an international business and tec...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>Dry Bulk Shipping Data Analyst</td>\n",
       "      <td>Permanent, Full Time</td>\n",
       "      <td>IHS GLOBAL PTE. LTD.</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Others</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>perform tasks and responsibilities as assigne...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>Senior Data Analyst</td>\n",
       "      <td>Permanent, Full Time</td>\n",
       "      <td>GRABTAXI HOLDINGS PTE. LTD.</td>\n",
       "      <td>OUE DOWNTOWN, 6 SHENTON WAY 068809</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>get to know the role: to design, set up, and m...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1354</th>\n",
       "      <td>Robotics and Automation Trainee (Attach and Tr...</td>\n",
       "      <td>Contract</td>\n",
       "      <td>PBA SYSTEMS PTE LTD</td>\n",
       "      <td>PBA BUILDING, 505 YISHUN INDUSTRIAL PARK A 768733</td>\n",
       "      <td>Engineering</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>switching careers? looking to contribute in a ...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1355</th>\n",
       "      <td>Software Engineer</td>\n",
       "      <td>Permanent</td>\n",
       "      <td>SENSETIME INTERNATIONAL PTE. LTD.</td>\n",
       "      <td>THE PLAZA, 7500A BEACH ROAD 199591</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>key responsibilities:  develop, build and impl...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1356</th>\n",
       "      <td>Senior  /  R&amp;D Engineer (AI, Deep Learning, Co...</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>PANASONIC R&amp;D CENTER SINGAPORE</td>\n",
       "      <td>202 BEDOK SOUTH AVENUE 1 469332</td>\n",
       "      <td>Sciences / Laboratory / R&amp;D</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>pan...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1357</th>\n",
       "      <td>IT Deep Learning Research Engineer</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>Company Undisclosed</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>we are looking for engineers to bring next gen...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1358</th>\n",
       "      <td>IT Research Scientist (Generative Models)</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>Company Undisclosed</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>the ai research group in the office of the cto...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1359</th>\n",
       "      <td>IT Computer Vision</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>Company Undisclosed</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>the position focuses on research and developme...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1360</th>\n",
       "      <td>Software Engineer  /  R&amp;D on Image / Camera pr...</td>\n",
       "      <td>Permanent</td>\n",
       "      <td>THE SUPREME HR ADVISORY PTE. LTD.</td>\n",
       "      <td>AZ @ PAYA LEBAR, 140 PAYA LEBAR ROAD 409015</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>experience with dft (discrete fourier transfo...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1361</th>\n",
       "      <td>Automation Developer  /  Bot Developer</td>\n",
       "      <td>Permanent, Full Time</td>\n",
       "      <td>KEYREPLY PTE. LTD.</td>\n",
       "      <td>NaN</td>\n",
       "      <td>General Management</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>you are: the head honcho who makes things happ...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1362</th>\n",
       "      <td>RPA Analyst</td>\n",
       "      <td>Contract</td>\n",
       "      <td>TERRABIT CONSULTING PTE. LTD.</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>to work with key stakeholders to execute enga...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1363</th>\n",
       "      <td>Research Assistant</td>\n",
       "      <td>Contract, Full Time</td>\n",
       "      <td>Company Undisclosed</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Others, Sciences / Laboratory / R&amp;D</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>the candidate willperform big data analysis an...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1364</th>\n",
       "      <td>Software Engineer</td>\n",
       "      <td>Permanent, Full Time</td>\n",
       "      <td>TERALYTICS PTE. LTD.</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Engineering</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>your role we are searching for highly-motivate...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1365</th>\n",
       "      <td>Full Stack Developer</td>\n",
       "      <td>Permanent, Full Time</td>\n",
       "      <td>ENT-VISION PTE. LTD.</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>front-end ux and dashboard development develo...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1366</th>\n",
       "      <td>Engineering, Post Execution Securities, Prime ...</td>\n",
       "      <td>Permanent, Full Time</td>\n",
       "      <td>GOLDMAN SACHS FUTURES PTE LTD</td>\n",
       "      <td>ONE RAFFLES LINK, 1 RAFFLES LINK 039393</td>\n",
       "      <td>Banking and Finance</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>what we do at goldman sachs, our engineers don...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1367</th>\n",
       "      <td>Engineering, Prime Services Technology, Prime ...</td>\n",
       "      <td>Permanent, Full Time</td>\n",
       "      <td>GOLDMAN SACHS FUTURES PTE LTD</td>\n",
       "      <td>ONE RAFFLES LINK, 1 RAFFLES LINK 039393</td>\n",
       "      <td>Banking and Finance</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>what we do at goldman sachs, our engineers don...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1368</th>\n",
       "      <td>Deployment and Runtime - Application Infrastru...</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>GOLDMAN SACHS SERVICES (SINGAPORE) PTE. LTD.</td>\n",
       "      <td>ONE RAFFLES LINK, 1 RAFFLES LINK 039393</td>\n",
       "      <td>Banking and Finance</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>what we do at goldman sachs, our engineers don...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1369</th>\n",
       "      <td>Deployment and Runtime - Application Infrastru...</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>GOLDMAN SACHS SERVICES (SINGAPORE) PTE. LTD.</td>\n",
       "      <td>ONE RAFFLES LINK, 1 RAFFLES LINK 039393</td>\n",
       "      <td>Banking and Finance</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>what we do at goldman sachs, our engineers don...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1370</th>\n",
       "      <td>Deployment and Runtime - Application Infrastru...</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>GOLDMAN SACHS SERVICES (SINGAPORE) PTE. LTD.</td>\n",
       "      <td>ONE RAFFLES LINK, 1 RAFFLES LINK 039393</td>\n",
       "      <td>Banking and Finance</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>what we do at goldman sachs, our engineers don...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1371</th>\n",
       "      <td>Head of Data Science, Google Premium Services ...</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>GOOGLE ASIA PACIFIC PTE. LTD.</td>\n",
       "      <td>MARINA BAY FINANCIAL CENTRE, 8 MARINA BOULEVAR...</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>company overview: google is not a conventional...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1372</th>\n",
       "      <td>Software Developer - Geospatial Technology Exp...</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>GOVERNMENT TECHNOLOGY AGENCY</td>\n",
       "      <td>MAPLETREE BUSINESS CITY, 10 PASIR PANJANG ROAD...</td>\n",
       "      <td>Information Technology, Public / Civil Service</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>the government technology agency (govtech) aim...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1373</th>\n",
       "      <td>DevOps Engineer, Government Digital Services</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>GOVERNMENT TECHNOLOGY AGENCY</td>\n",
       "      <td>MAPLETREE BUSINESS CITY, 10 PASIR PANJANG ROAD...</td>\n",
       "      <td>Information Technology, Public / Civil Service</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>the government digital services (gds) team aim...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1374</th>\n",
       "      <td>Advanced Software Engineer</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>HOMAGE CO PTE. LTD.</td>\n",
       "      <td>157B ROCHOR ROAD 188432</td>\n",
       "      <td>Engineering, Information Technology</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>engineers at homage build things that matter –...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1375</th>\n",
       "      <td>Software Engineer - Singapore</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>GOOGLE ASIA PACIFIC PTE. LTD.</td>\n",
       "      <td>MARINA BAY FINANCIAL CENTRE, 8 MARINA BOULEVAR...</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>company overview: google is not a conventional...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1376</th>\n",
       "      <td>Python Developer</td>\n",
       "      <td>Permanent, Full Time</td>\n",
       "      <td>INFOSYS LIMITED SINGAPORE BRANCH</td>\n",
       "      <td>SINGAPORE LAND TOWER, 50 RAFFLES PLACE 048623</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>we are building a scalable multithreaded natur...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1377</th>\n",
       "      <td>NLP Engineer</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>TAIGER SINGAPORE PTE. LTD.</td>\n",
       "      <td>26B TEMPLE STREET 058571</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>taigerseeks a nlp engineer to join our interna...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1378</th>\n",
       "      <td>Research Scientist (1-year contract), Fujitsu-...</td>\n",
       "      <td>Contract</td>\n",
       "      <td>SINGAPORE MANAGEMENT UNIVERSITY</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Education and Training</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>company description  singapore management uni...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1379</th>\n",
       "      <td>Study Manager</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>MANPOWER STAFFING SERVICES (SINGAPORE) PTE LTD</td>\n",
       "      <td>GUOCO TOWER, 1 WALLICH STREET 078881</td>\n",
       "      <td>Sciences / Laboratory / R&amp;D</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>job specification  line manage a metabolism te...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1380</th>\n",
       "      <td>Research Scientist (Computer Vision)</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>NEURONCREDIT PTE. LTD.</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>this research scientist will work with our hi...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1381</th>\n",
       "      <td>Research Scientist</td>\n",
       "      <td>Contract, Full Time</td>\n",
       "      <td>NANYANG TECHNOLOGICAL UNIVERSITY</td>\n",
       "      <td>NANYANG TECHNOLOGICAL UNIVERSITY, 50 NANYANG A...</td>\n",
       "      <td>Engineering, Sciences / Laboratory / R&amp;D</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>the physical analysis and cryptographic engine...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1382</th>\n",
       "      <td>Research Scientist (CS), IHPC</td>\n",
       "      <td>Contract, Full Time</td>\n",
       "      <td>A*STAR RESEARCH ENTITIES</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Sciences / Laboratory / R&amp;D</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>he/she will mainly work on data analysis, mode...</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1383</th>\n",
       "      <td>Research Scientist (Biomedical Science)</td>\n",
       "      <td>Full Time</td>\n",
       "      <td>IMMUNOSCAPE PTE. LTD.</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Sciences / Laboratory / R&amp;D</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>immunoscape is looking for a highly motivated ...</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1384 rows × 22 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                              Job Title              Job Type  \\\n",
       "0                                          Data Analyst             Full Time   \n",
       "1                                          Data Analyst             Permanent   \n",
       "2                              Datawarehouse Consultant             Full Time   \n",
       "3                                          Data analyst             Full Time   \n",
       "4                                          Data Analyst             Full Time   \n",
       "5        Senior Associate, Data and Analytics, Advisory             Full Time   \n",
       "6                Associate, Advisory Data and Analytics             Full Time   \n",
       "7                                          Data Analyst              Contract   \n",
       "8                                 Senior Data Scientist             Full Time   \n",
       "9                  Data Analyst, Transformation Project              Contract   \n",
       "10    Commercial Data Analyst (East, Advanced Excel,...  Permanent, Full Time   \n",
       "11                                  Senior Data Analyst  Permanent, Full Time   \n",
       "12                      Data Analyst (SAS, R or Python)              Contract   \n",
       "13                                         Data Analyst             Full Time   \n",
       "14                                         Data Analyst              Contract   \n",
       "15                                Business Data Analyst   Permanent, Contract   \n",
       "16                                         Data Analyst             Full Time   \n",
       "17                                  EBS Spark Developer             Full Time   \n",
       "18    Wealth Management – Senior Data Analyst, Clien...  Permanent, Full Time   \n",
       "19                                        DATA  ANALYST             Full Time   \n",
       "20      Business Analyst Project Manager - Data Science   Permanent, Contract   \n",
       "21          Global CDS Operations OneTouch Data Analyst             Full Time   \n",
       "22                              Data Management Analyst   Contract, Full Time   \n",
       "23                                         Data Analyst             Permanent   \n",
       "24     Senior / Data Analyst (1 Year Contract) - S$3000              Contract   \n",
       "25    Data Analyst (Stack) - 1 year contract, Group ...             Full Time   \n",
       "26                         Senior Analyst, Data Science             Full Time   \n",
       "27                                Informatica Developer             Permanent   \n",
       "28                       Dry Bulk Shipping Data Analyst  Permanent, Full Time   \n",
       "29                                  Senior Data Analyst  Permanent, Full Time   \n",
       "...                                                 ...                   ...   \n",
       "1354  Robotics and Automation Trainee (Attach and Tr...              Contract   \n",
       "1355                                  Software Engineer             Permanent   \n",
       "1356  Senior  /  R&D Engineer (AI, Deep Learning, Co...             Full Time   \n",
       "1357                 IT Deep Learning Research Engineer             Full Time   \n",
       "1358          IT Research Scientist (Generative Models)             Full Time   \n",
       "1359                                 IT Computer Vision             Full Time   \n",
       "1360  Software Engineer  /  R&D on Image / Camera pr...             Permanent   \n",
       "1361             Automation Developer  /  Bot Developer  Permanent, Full Time   \n",
       "1362                                        RPA Analyst              Contract   \n",
       "1363                                 Research Assistant   Contract, Full Time   \n",
       "1364                                  Software Engineer  Permanent, Full Time   \n",
       "1365                               Full Stack Developer  Permanent, Full Time   \n",
       "1366  Engineering, Post Execution Securities, Prime ...  Permanent, Full Time   \n",
       "1367  Engineering, Prime Services Technology, Prime ...  Permanent, Full Time   \n",
       "1368  Deployment and Runtime - Application Infrastru...             Full Time   \n",
       "1369  Deployment and Runtime - Application Infrastru...             Full Time   \n",
       "1370  Deployment and Runtime - Application Infrastru...             Full Time   \n",
       "1371  Head of Data Science, Google Premium Services ...             Full Time   \n",
       "1372  Software Developer - Geospatial Technology Exp...             Full Time   \n",
       "1373       DevOps Engineer, Government Digital Services             Full Time   \n",
       "1374                         Advanced Software Engineer             Full Time   \n",
       "1375                      Software Engineer - Singapore             Full Time   \n",
       "1376                                   Python Developer  Permanent, Full Time   \n",
       "1377                                       NLP Engineer             Full Time   \n",
       "1378  Research Scientist (1-year contract), Fujitsu-...              Contract   \n",
       "1379                                      Study Manager             Full Time   \n",
       "1380               Research Scientist (Computer Vision)             Full Time   \n",
       "1381                                 Research Scientist   Contract, Full Time   \n",
       "1382                      Research Scientist (CS), IHPC   Contract, Full Time   \n",
       "1383            Research Scientist (Biomedical Science)             Full Time   \n",
       "\n",
       "                                         Company Name  \\\n",
       "0     INTERNATIONAL SOS TECHNOLOGY SERVICES PTE. LTD.   \n",
       "1                                  ASCENTIS PTE. LTD.   \n",
       "2                   R SYSTEMS (SINGAPORE) PTE LIMITED   \n",
       "3                                TRAKOMATIC PTE. LTD.   \n",
       "4                                 GUMI ASIA PTE. LTD.   \n",
       "5                    ERNST & YOUNG ADVISORY PTE. LTD.   \n",
       "6                    ERNST & YOUNG ADVISORY PTE. LTD.   \n",
       "7                      FUTUREX TECHNOLOGIES PTE. LTD.   \n",
       "8                             SINGAPORE POWER LIMITED   \n",
       "9                    CHANDLER MACLEOD GROUP PTE. LTD.   \n",
       "10               GMP RECRUITMENT SERVICES (S) PTE LTD   \n",
       "11                                       99 PTE. LTD.   \n",
       "12              OPTIMUM SOLUTIONS (SINGAPORE) PTE LTD   \n",
       "13                MINDTREE LIMITED (SINGAPORE BRANCH)   \n",
       "14            HYDROGEN CONSULTING SOLUTIONS PTE. LTD.   \n",
       "15           RGF TALENT SOLUTIONS SINGAPORE PTE. LTD.   \n",
       "16                                Company Undisclosed   \n",
       "17                      CITTABASE SOLUTIONS PTE. LTD.   \n",
       "18                   DEUTSCHE BANK AKTIENGESELLSCHAFT   \n",
       "19                                Company Undisclosed   \n",
       "20                                Company Undisclosed   \n",
       "21                  BCD TRAVEL ASIA PACIFIC PTE. LTD.   \n",
       "22     MANPOWER STAFFING SERVICES (SINGAPORE) PTE LTD   \n",
       "23                    ABAKUS (ASIA PACIFIC) PTE. LTD.   \n",
       "24                        CLARIST RESOURCES PTE. LTD.   \n",
       "25        OVERSEA-CHINESE BANKING CORPORATION LIMITED   \n",
       "26                   LAZADA SOUTH EAST ASIA PTE. LTD.   \n",
       "27                                      ADDSTONES SAS   \n",
       "28                               IHS GLOBAL PTE. LTD.   \n",
       "29                        GRABTAXI HOLDINGS PTE. LTD.   \n",
       "...                                               ...   \n",
       "1354                              PBA SYSTEMS PTE LTD   \n",
       "1355                SENSETIME INTERNATIONAL PTE. LTD.   \n",
       "1356                   PANASONIC R&D CENTER SINGAPORE   \n",
       "1357                              Company Undisclosed   \n",
       "1358                              Company Undisclosed   \n",
       "1359                              Company Undisclosed   \n",
       "1360                THE SUPREME HR ADVISORY PTE. LTD.   \n",
       "1361                               KEYREPLY PTE. LTD.   \n",
       "1362                    TERRABIT CONSULTING PTE. LTD.   \n",
       "1363                              Company Undisclosed   \n",
       "1364                             TERALYTICS PTE. LTD.   \n",
       "1365                             ENT-VISION PTE. LTD.   \n",
       "1366                    GOLDMAN SACHS FUTURES PTE LTD   \n",
       "1367                    GOLDMAN SACHS FUTURES PTE LTD   \n",
       "1368     GOLDMAN SACHS SERVICES (SINGAPORE) PTE. LTD.   \n",
       "1369     GOLDMAN SACHS SERVICES (SINGAPORE) PTE. LTD.   \n",
       "1370     GOLDMAN SACHS SERVICES (SINGAPORE) PTE. LTD.   \n",
       "1371                    GOOGLE ASIA PACIFIC PTE. LTD.   \n",
       "1372                     GOVERNMENT TECHNOLOGY AGENCY   \n",
       "1373                     GOVERNMENT TECHNOLOGY AGENCY   \n",
       "1374                              HOMAGE CO PTE. LTD.   \n",
       "1375                    GOOGLE ASIA PACIFIC PTE. LTD.   \n",
       "1376                 INFOSYS LIMITED SINGAPORE BRANCH   \n",
       "1377                       TAIGER SINGAPORE PTE. LTD.   \n",
       "1378                  SINGAPORE MANAGEMENT UNIVERSITY   \n",
       "1379   MANPOWER STAFFING SERVICES (SINGAPORE) PTE LTD   \n",
       "1380                           NEURONCREDIT PTE. LTD.   \n",
       "1381                 NANYANG TECHNOLOGICAL UNIVERSITY   \n",
       "1382                         A*STAR RESEARCH ENTITIES   \n",
       "1383                            IMMUNOSCAPE PTE. LTD.   \n",
       "\n",
       "                                           Job Location  \\\n",
       "0            ODEON TOWERS, 331 NORTH BRIDGE ROAD 188720   \n",
       "1                               3015A UBI ROAD 1 408705   \n",
       "2              HOI HUP BUILDING, 16 JALAN KILANG 159416   \n",
       "3              ALICE@MEDIAPOLIS, 29 MEDIA CIRCLE 138565   \n",
       "4                   CONNEXIS, 1 FUSIONOPOLIS WAY 138632   \n",
       "5                                 1 RAFFLES QUAY 048583   \n",
       "6                                 1 RAFFLES QUAY 048583   \n",
       "7     GERMAN CENTRE, 25 INTERNATIONAL BUSINESS PARK ...   \n",
       "8            SP GROUP BUILDING, 2 KALLANG SECTOR 349277   \n",
       "9              KECK SENG TOWER, 133 CECIL STREET 069535   \n",
       "10        ONE FINLAYSON GREEN, 1 FINLAYSON GREEN 049246   \n",
       "11                        71 AYER RAJAH CRESCENT 139951   \n",
       "12    PLAZA 8 @ CBP, 1 CHANGI BUSINESS PARK CRESCENT...   \n",
       "13    HONEYWELL BUILDING, 17 CHANGI BUSINESS PARK CE...   \n",
       "14             PRUDENTIAL TOWER, 30 CECIL STREET 049712   \n",
       "15                              71 ROBINSON ROAD 068895   \n",
       "16                                                  NaN   \n",
       "17                                                  NaN   \n",
       "18                                1 RAFFLES QUAY 048583   \n",
       "19                                                  NaN   \n",
       "20          HONG LEONG BUILDING, 16 RAFFLES QUAY 048581   \n",
       "21                   TWENTY ANSON, 20 ANSON ROAD 079912   \n",
       "22                 GUOCO TOWER, 1 WALLICH STREET 078881   \n",
       "23                                                  NaN   \n",
       "24              ORCHARD TOWERS, 400 ORCHARD ROAD 238875   \n",
       "25                              63 CHULIA STREET 049514   \n",
       "26                      AXA TOWER, 8 SHENTON WAY 068811   \n",
       "27                                                  NaN   \n",
       "28                                                  NaN   \n",
       "29                   OUE DOWNTOWN, 6 SHENTON WAY 068809   \n",
       "...                                                 ...   \n",
       "1354  PBA BUILDING, 505 YISHUN INDUSTRIAL PARK A 768733   \n",
       "1355                 THE PLAZA, 7500A BEACH ROAD 199591   \n",
       "1356                    202 BEDOK SOUTH AVENUE 1 469332   \n",
       "1357                                                NaN   \n",
       "1358                                                NaN   \n",
       "1359                                                NaN   \n",
       "1360        AZ @ PAYA LEBAR, 140 PAYA LEBAR ROAD 409015   \n",
       "1361                                                NaN   \n",
       "1362                                                NaN   \n",
       "1363                                                NaN   \n",
       "1364                                                NaN   \n",
       "1365                                                NaN   \n",
       "1366            ONE RAFFLES LINK, 1 RAFFLES LINK 039393   \n",
       "1367            ONE RAFFLES LINK, 1 RAFFLES LINK 039393   \n",
       "1368            ONE RAFFLES LINK, 1 RAFFLES LINK 039393   \n",
       "1369            ONE RAFFLES LINK, 1 RAFFLES LINK 039393   \n",
       "1370            ONE RAFFLES LINK, 1 RAFFLES LINK 039393   \n",
       "1371  MARINA BAY FINANCIAL CENTRE, 8 MARINA BOULEVAR...   \n",
       "1372  MAPLETREE BUSINESS CITY, 10 PASIR PANJANG ROAD...   \n",
       "1373  MAPLETREE BUSINESS CITY, 10 PASIR PANJANG ROAD...   \n",
       "1374                            157B ROCHOR ROAD 188432   \n",
       "1375  MARINA BAY FINANCIAL CENTRE, 8 MARINA BOULEVAR...   \n",
       "1376      SINGAPORE LAND TOWER, 50 RAFFLES PLACE 048623   \n",
       "1377                           26B TEMPLE STREET 058571   \n",
       "1378                                                NaN   \n",
       "1379               GUOCO TOWER, 1 WALLICH STREET 078881   \n",
       "1380                                                NaN   \n",
       "1381  NANYANG TECHNOLOGICAL UNIVERSITY, 50 NANYANG A...   \n",
       "1382                                                NaN   \n",
       "1383                                                NaN   \n",
       "\n",
       "                                           Job Industry  Contract  Internship  \\\n",
       "0                                Information Technology         0           0   \n",
       "1                                Information Technology         0           0   \n",
       "2                                Information Technology         0           0   \n",
       "3                                Information Technology         0           0   \n",
       "4                                                Others         0           0   \n",
       "5                                           Consulting          0           0   \n",
       "6                                           Consulting          0           0   \n",
       "7                                Information Technology         1           0   \n",
       "8                                Information Technology         0           0   \n",
       "9                                Information Technology         1           0   \n",
       "10                                  Banking and Finance         0           0   \n",
       "11                                          Engineering         0           0   \n",
       "12                               Information Technology         1           0   \n",
       "13                               Information Technology         0           0   \n",
       "14                                  Banking and Finance         1           0   \n",
       "15                               Information Technology         1           0   \n",
       "16                                               Others         0           0   \n",
       "17                               Information Technology         0           0   \n",
       "18                                  Banking and Finance         0           0   \n",
       "19                               Information Technology         0           0   \n",
       "20                               Information Technology         1           0   \n",
       "21                                     Travel / Tourism         0           0   \n",
       "22               Advertising / Media , Customer Service         1           0   \n",
       "23                               Information Technology         0           0   \n",
       "24    Information Technology, Logistics / Supply Cha...         1           0   \n",
       "25                                               Others         0           0   \n",
       "26                               Information Technology         0           0   \n",
       "27    Consulting , Banking and Finance, Information ...         0           0   \n",
       "28                                               Others         0           0   \n",
       "29                               Information Technology         0           0   \n",
       "...                                                 ...       ...         ...   \n",
       "1354                                        Engineering         1           0   \n",
       "1355                             Information Technology         0           0   \n",
       "1356                        Sciences / Laboratory / R&D         0           0   \n",
       "1357                             Information Technology         0           0   \n",
       "1358                             Information Technology         0           0   \n",
       "1359                             Information Technology         0           0   \n",
       "1360                             Information Technology         0           0   \n",
       "1361                                 General Management         0           0   \n",
       "1362                             Information Technology         1           0   \n",
       "1363                Others, Sciences / Laboratory / R&D         1           0   \n",
       "1364                                        Engineering         0           0   \n",
       "1365                             Information Technology         0           0   \n",
       "1366                                Banking and Finance         0           0   \n",
       "1367                                Banking and Finance         0           0   \n",
       "1368                                Banking and Finance         0           0   \n",
       "1369                                Banking and Finance         0           0   \n",
       "1370                                Banking and Finance         0           0   \n",
       "1371                             Information Technology         0           0   \n",
       "1372     Information Technology, Public / Civil Service         0           0   \n",
       "1373     Information Technology, Public / Civil Service         0           0   \n",
       "1374                Engineering, Information Technology         0           0   \n",
       "1375                             Information Technology         0           0   \n",
       "1376                             Information Technology         0           0   \n",
       "1377                             Information Technology         0           0   \n",
       "1378                             Education and Training         1           0   \n",
       "1379                        Sciences / Laboratory / R&D         0           0   \n",
       "1380                             Information Technology         0           0   \n",
       "1381           Engineering, Sciences / Laboratory / R&D         1           0   \n",
       "1382                        Sciences / Laboratory / R&D         1           0   \n",
       "1383                        Sciences / Laboratory / R&D         0           0   \n",
       "\n",
       "      Temporary  Permanent                                    Job Description  \\\n",
       "0             0          0  in our digital innovation services centre (dis...   \n",
       "1             0          1  data analyst job description: the role will be...   \n",
       "2             0          0   experience working with business users to con...   \n",
       "3             0          0  the retail data analyst will lead and perform ...   \n",
       "4             0          0   perform evaluations and problem investigation...   \n",
       "5             0          0  powered by big data and advanced technologies,...   \n",
       "6             0          0  powered by big data and advanced technologies,...   \n",
       "7             0          0  the successful candidate will be responsible f...   \n",
       "8             0          0  why work for us we power the nation.  make the...   \n",
       "9             0          0   23 months contract global bank, based in cbd ...   \n",
       "10            0          1   a well-established fmcg.  responsibilities:​​...   \n",
       "11            0          1  99.co is looking for a senior data analyst to ...   \n",
       "12            0          0  optimum solutions (company registration number...   \n",
       "13            0          0  1) around 6+ years of it experience across var...   \n",
       "14            0          0  min. 2 years’ experience in banking industry p...   \n",
       "15            0          1   primarily support the global marketing team w...   \n",
       "16            0          0   develop and implement data collection, data a...   \n",
       "17            0          0  oracle ebs data warehousing reporting and anal...   \n",
       "18            0          1  private & commercial bank deutsche bank's newl...   \n",
       "19            0          0   identify valuable data sources and automate c...   \n",
       "20            0          1  who we are:  we ns solutions asia pacific (nsa...   \n",
       "21            0          0  where willyourcareer take you?     we’re not j...   \n",
       "22            0          0  we are looking for new members to help us achi...   \n",
       "23            0          1  to reinvent an industry, you need to build an ...   \n",
       "24            0          0  our client for this role well-established, hom...   \n",
       "25            0          0  job description:   work alongside the group cu...   \n",
       "26            0          0  team introduction lazada is the number one onl...   \n",
       "27            0          1  gfi group is an international business and tec...   \n",
       "28            0          1   perform tasks and responsibilities as assigne...   \n",
       "29            0          1  get to know the role: to design, set up, and m...   \n",
       "...         ...        ...                                                ...   \n",
       "1354          0          0  switching careers? looking to contribute in a ...   \n",
       "1355          0          1  key responsibilities:  develop, build and impl...   \n",
       "1356          0          0                                             pan...   \n",
       "1357          0          0  we are looking for engineers to bring next gen...   \n",
       "1358          0          0  the ai research group in the office of the cto...   \n",
       "1359          0          0  the position focuses on research and developme...   \n",
       "1360          0          1   experience with dft (discrete fourier transfo...   \n",
       "1361          0          1  you are: the head honcho who makes things happ...   \n",
       "1362          0          0   to work with key stakeholders to execute enga...   \n",
       "1363          0          0  the candidate willperform big data analysis an...   \n",
       "1364          0          1  your role we are searching for highly-motivate...   \n",
       "1365          0          1   front-end ux and dashboard development develo...   \n",
       "1366          0          1  what we do at goldman sachs, our engineers don...   \n",
       "1367          0          1  what we do at goldman sachs, our engineers don...   \n",
       "1368          0          0  what we do at goldman sachs, our engineers don...   \n",
       "1369          0          0  what we do at goldman sachs, our engineers don...   \n",
       "1370          0          0  what we do at goldman sachs, our engineers don...   \n",
       "1371          0          0  company overview: google is not a conventional...   \n",
       "1372          0          0  the government technology agency (govtech) aim...   \n",
       "1373          0          0  the government digital services (gds) team aim...   \n",
       "1374          0          0  engineers at homage build things that matter –...   \n",
       "1375          0          0  company overview: google is not a conventional...   \n",
       "1376          0          1  we are building a scalable multithreaded natur...   \n",
       "1377          0          0  taigerseeks a nlp engineer to join our interna...   \n",
       "1378          0          0   company description  singapore management uni...   \n",
       "1379          0          0  job specification  line manage a metabolism te...   \n",
       "1380          0          0   this research scientist will work with our hi...   \n",
       "1381          0          0  the physical analysis and cryptographic engine...   \n",
       "1382          0          0  he/she will mainly work on data analysis, mode...   \n",
       "1383          0          0  immunoscape is looking for a highly motivated ...   \n",
       "\n",
       "            ...         Salary_Category  Executive  Fresh/entry level  \\\n",
       "0           ...                       1        0.0                0.0   \n",
       "1           ...                       0        0.0                0.0   \n",
       "2           ...                       1        1.0                0.0   \n",
       "3           ...                       0        0.0                0.0   \n",
       "4           ...                       0        0.0                0.0   \n",
       "5           ...                       1        1.0                0.0   \n",
       "6           ...                       0        1.0                0.0   \n",
       "7           ...                       1        1.0                0.0   \n",
       "8           ...                       1        0.0                0.0   \n",
       "9           ...                       0        0.0                0.0   \n",
       "10          ...                       0        1.0                0.0   \n",
       "11          ...                       0        1.0                0.0   \n",
       "12          ...                       1        0.0                0.0   \n",
       "13          ...                       1        0.0                0.0   \n",
       "14          ...                       0        0.0                0.0   \n",
       "15          ...                       1        0.0                0.0   \n",
       "16          ...                       1        0.0                0.0   \n",
       "17          ...                       1        1.0                0.0   \n",
       "18          ...                       1        0.0                0.0   \n",
       "19          ...                       1        0.0                0.0   \n",
       "20          ...                       1        0.0                0.0   \n",
       "21          ...                       0        0.0                0.0   \n",
       "22          ...                       0        0.0                0.0   \n",
       "23          ...                       0        1.0                0.0   \n",
       "24          ...                       0        1.0                0.0   \n",
       "25          ...                       1        0.0                0.0   \n",
       "26          ...                       0        0.0                0.0   \n",
       "27          ...                       1        0.0                0.0   \n",
       "28          ...                       0        1.0                0.0   \n",
       "29          ...                       1        1.0                0.0   \n",
       "...         ...                     ...        ...                ...   \n",
       "1354        ...                       0        0.0                0.0   \n",
       "1355        ...                       0        1.0                0.0   \n",
       "1356        ...                       0        1.0                0.0   \n",
       "1357        ...                       1        0.0                0.0   \n",
       "1358        ...                       1        0.0                0.0   \n",
       "1359        ...                       1        0.0                0.0   \n",
       "1360        ...                       0        1.0                0.0   \n",
       "1361        ...                       0        0.0                0.0   \n",
       "1362        ...                       1        0.0                0.0   \n",
       "1363        ...                       0        0.0                0.0   \n",
       "1364        ...                       1        0.0                0.0   \n",
       "1365        ...                       0        1.0                0.0   \n",
       "1366        ...                       1        1.0                0.0   \n",
       "1367        ...                       1        1.0                0.0   \n",
       "1368        ...                       1        1.0                0.0   \n",
       "1369        ...                       1        1.0                0.0   \n",
       "1370        ...                       1        1.0                0.0   \n",
       "1371        ...                       1        0.0                0.0   \n",
       "1372        ...                       0        0.0                0.0   \n",
       "1373        ...                       1        0.0                0.0   \n",
       "1374        ...                       1        0.0                0.0   \n",
       "1375        ...                       1        1.0                0.0   \n",
       "1376        ...                       0        0.0                0.0   \n",
       "1377        ...                       0        1.0                0.0   \n",
       "1378        ...                       0        0.0                0.0   \n",
       "1379        ...                       1        0.0                0.0   \n",
       "1380        ...                       1        1.0                0.0   \n",
       "1381        ...                       0        0.0                0.0   \n",
       "1382        ...                       1        0.0                0.0   \n",
       "1383        ...                       0        0.0                0.0   \n",
       "\n",
       "      Junior Executive  Manager  Middle Management  Non-executive  \\\n",
       "0                  0.0      0.0                0.0            0.0   \n",
       "1                  0.0      0.0                0.0            0.0   \n",
       "2                  0.0      0.0                0.0            0.0   \n",
       "3                  0.0      0.0                0.0            0.0   \n",
       "4                  1.0      0.0                0.0            0.0   \n",
       "5                  0.0      0.0                0.0            0.0   \n",
       "6                  0.0      0.0                0.0            0.0   \n",
       "7                  0.0      0.0                0.0            0.0   \n",
       "8                  0.0      0.0                1.0            0.0   \n",
       "9                  0.0      0.0                0.0            0.0   \n",
       "10                 0.0      0.0                0.0            0.0   \n",
       "11                 0.0      0.0                0.0            0.0   \n",
       "12                 0.0      1.0                0.0            0.0   \n",
       "13                 0.0      0.0                0.0            0.0   \n",
       "14                 0.0      0.0                0.0            0.0   \n",
       "15                 0.0      0.0                0.0            0.0   \n",
       "16                 0.0      0.0                0.0            0.0   \n",
       "17                 0.0      0.0                0.0            0.0   \n",
       "18                 0.0      0.0                0.0            0.0   \n",
       "19                 0.0      0.0                0.0            0.0   \n",
       "20                 0.0      1.0                0.0            0.0   \n",
       "21                 1.0      0.0                0.0            1.0   \n",
       "22                 0.0      0.0                0.0            1.0   \n",
       "23                 1.0      0.0                0.0            0.0   \n",
       "24                 0.0      0.0                0.0            0.0   \n",
       "25                 0.0      0.0                0.0            0.0   \n",
       "26                 0.0      0.0                0.0            0.0   \n",
       "27                 0.0      1.0                0.0            0.0   \n",
       "28                 0.0      0.0                0.0            0.0   \n",
       "29                 0.0      0.0                0.0            0.0   \n",
       "...                ...      ...                ...            ...   \n",
       "1354               0.0      0.0                0.0            0.0   \n",
       "1355               1.0      0.0                0.0            0.0   \n",
       "1356               0.0      0.0                0.0            0.0   \n",
       "1357               0.0      0.0                0.0            0.0   \n",
       "1358               0.0      0.0                0.0            0.0   \n",
       "1359               0.0      0.0                0.0            0.0   \n",
       "1360               0.0      0.0                0.0            0.0   \n",
       "1361               1.0      0.0                0.0            0.0   \n",
       "1362               0.0      0.0                1.0            0.0   \n",
       "1363               0.0      0.0                0.0            0.0   \n",
       "1364               0.0      0.0                0.0            0.0   \n",
       "1365               0.0      0.0                0.0            0.0   \n",
       "1366               0.0      0.0                0.0            0.0   \n",
       "1367               0.0      0.0                0.0            0.0   \n",
       "1368               0.0      0.0                0.0            0.0   \n",
       "1369               0.0      0.0                0.0            0.0   \n",
       "1370               0.0      0.0                0.0            0.0   \n",
       "1371               0.0      0.0                0.0            0.0   \n",
       "1372               0.0      0.0                1.0            0.0   \n",
       "1373               0.0      0.0                1.0            0.0   \n",
       "1374               0.0      0.0                0.0            0.0   \n",
       "1375               0.0      0.0                0.0            0.0   \n",
       "1376               0.0      0.0                0.0            0.0   \n",
       "1377               0.0      0.0                0.0            0.0   \n",
       "1378               0.0      1.0                0.0            0.0   \n",
       "1379               0.0      1.0                0.0            0.0   \n",
       "1380               0.0      0.0                0.0            0.0   \n",
       "1381               0.0      0.0                0.0            1.0   \n",
       "1382               0.0      0.0                0.0            1.0   \n",
       "1383               0.0      0.0                0.0            0.0   \n",
       "\n",
       "      Professional  Senior Executive  Senior Management  \n",
       "0              1.0               0.0                0.0  \n",
       "1              0.0               1.0                0.0  \n",
       "2              0.0               0.0                0.0  \n",
       "3              1.0               0.0                0.0  \n",
       "4              0.0               0.0                0.0  \n",
       "5              0.0               0.0                0.0  \n",
       "6              0.0               0.0                0.0  \n",
       "7              0.0               0.0                0.0  \n",
       "8              0.0               0.0                0.0  \n",
       "9              0.0               1.0                0.0  \n",
       "10             0.0               1.0                0.0  \n",
       "11             1.0               1.0                0.0  \n",
       "12             0.0               0.0                1.0  \n",
       "13             1.0               0.0                0.0  \n",
       "14             1.0               0.0                0.0  \n",
       "15             0.0               1.0                0.0  \n",
       "16             1.0               0.0                0.0  \n",
       "17             0.0               0.0                0.0  \n",
       "18             1.0               0.0                0.0  \n",
       "19             0.0               1.0                0.0  \n",
       "20             0.0               0.0                0.0  \n",
       "21             0.0               0.0                0.0  \n",
       "22             0.0               0.0                0.0  \n",
       "23             0.0               0.0                0.0  \n",
       "24             0.0               0.0                0.0  \n",
       "25             0.0               1.0                0.0  \n",
       "26             0.0               1.0                0.0  \n",
       "27             0.0               0.0                0.0  \n",
       "28             0.0               0.0                0.0  \n",
       "29             0.0               1.0                0.0  \n",
       "...            ...               ...                ...  \n",
       "1354           1.0               0.0                0.0  \n",
       "1355           0.0               1.0                0.0  \n",
       "1356           0.0               0.0                0.0  \n",
       "1357           1.0               0.0                0.0  \n",
       "1358           1.0               0.0                0.0  \n",
       "1359           1.0               0.0                0.0  \n",
       "1360           0.0               0.0                0.0  \n",
       "1361           0.0               0.0                0.0  \n",
       "1362           0.0               0.0                0.0  \n",
       "1363           1.0               0.0                0.0  \n",
       "1364           1.0               0.0                0.0  \n",
       "1365           0.0               0.0                0.0  \n",
       "1366           0.0               0.0                0.0  \n",
       "1367           0.0               0.0                0.0  \n",
       "1368           0.0               0.0                0.0  \n",
       "1369           0.0               0.0                0.0  \n",
       "1370           0.0               0.0                0.0  \n",
       "1371           0.0               0.0                1.0  \n",
       "1372           0.0               0.0                0.0  \n",
       "1373           0.0               0.0                0.0  \n",
       "1374           1.0               0.0                0.0  \n",
       "1375           0.0               0.0                0.0  \n",
       "1376           1.0               0.0                0.0  \n",
       "1377           0.0               0.0                0.0  \n",
       "1378           0.0               0.0                0.0  \n",
       "1379           0.0               0.0                0.0  \n",
       "1380           0.0               0.0                0.0  \n",
       "1381           1.0               0.0                0.0  \n",
       "1382           0.0               0.0                0.0  \n",
       "1383           1.0               0.0                0.0  \n",
       "\n",
       "[1384 rows x 22 columns]"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Dropping not required columns\n",
    "df.drop(columns=[\"Min Salary\", \"Max Salary\", \"Salary Paid\", \"Date Posted\", \"Job Level\"], inplace=True)\n",
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Dummy Coding Job Industries\n",
    "list2 = []\n",
    "# Split jobs with multiple industries into a single list of unique industry \n",
    "for x in df[\"Job Industry\"].unique():\n",
    "    for z in range(len(x.split(\", \"))):\n",
    "        if x.split(\", \")[z] not in list2:\n",
    "            list2.append(x.split(\", \")[z])   \n",
    "            \n",
    "# for each df rows, go through each job industry, flag 1  when found a matching industry else 0\n",
    "for count, x in enumerate(df[\"Job Industry\"].apply(lambda x: x.split(\", \"))):\n",
    "    for y in list2:\n",
    "        if y in x:\n",
    "            df.loc[count, y] = int(1)\n",
    "        else:\n",
    "            df.loc[count, y] = int(0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "------\n",
    "### Start of Modelling"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### QUESTION 1: Factors that impact salary\n",
    "- Finding out what factors impact salary, through a classification problem.\n",
    "- In this case, I will assign `Salary_Category` above the median as 1, `Salary_Category` below or equals the median as 0.\n",
    "- With the `Salary_Category` category, I will try to use selected features to identify if they impact salary.\n",
    "\n",
    "Out of the several features I have obtained, I will attempt to build simple models first before expanding with additional features to see if the additional features better predict `Salary_Category`\n",
    "\n",
    "For a start, I will use `Job Level` as factor that impacts salary."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "from nltk.corpus import stopwords\n",
    "from sklearn.model_selection import train_test_split, cross_val_score, GridSearchCV\n",
    "from sklearn.feature_extraction.text import CountVectorizer, TfidfVectorizer\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.naive_bayes import MultinomialNB\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.metrics import classification_report, confusion_matrix, accuracy_score\n",
    "from sklearn.metrics import roc_auc_score, roc_curve, auc\n",
    "from sklearn.ensemble import RandomForestClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Defining stop words for NLP\n",
    "stop = stopwords.words('english')\n",
    "stop += [\"data\"]  # to add on more stopwords"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Simple Case 1: Using `Job Levels` to see if it can predict `Salary_Category`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df[['Executive', 'Fresh/entry level', 'Junior Executive', 'Manager', 'Middle Management', 'Non-executive', \\\n",
    "         'Professional', 'Senior Executive', 'Senior Management']].sum().sort_values().plot.bar()\n",
    "ax = plt.gca()\n",
    "plt.title(\"Distribution of Job Types\")\n",
    "plt.ylabel(\"Counts\")\n",
    "plt.xticks()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Overall graph visually looks like certain features within the Job Types may be able to explain salary differences\n",
    "- I will investigate using a few models to see if the models used can predict the Salary category correctly."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Defining the features for Case 1\n",
    "features1 = ['Executive', 'Fresh/entry level', 'Junior Executive', 'Manager', 'Middle Management', 'Non-executive', 'Professional', 'Senior Executive', 'Senior Management']\n",
    "X1 = df[features1]\n",
    "y1 = df[\"Salary_Category\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train1, X_test1, y_train, y_test = train_test_split(X1, y1, test_size=0.3, random_state=91)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Baseline: 0.5173410404624278\n"
     ]
    }
   ],
   "source": [
    "# Baseline - Majority Class (0) - Indicates how likely to predict the correct value based on majority class randomly\n",
    "print(\"Baseline:\", len(df[df[\"Salary_Category\"] == 0])/len(df[\"Salary_Category\"]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Using Logistic Regression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'C': 0.01, 'penalty': 'l2'}\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "Below Median       0.76      0.65      0.70       219\n",
      "Above Median       0.66      0.77      0.71       197\n",
      "\n",
      "   micro avg       0.70      0.70      0.70       416\n",
      "   macro avg       0.71      0.71      0.70       416\n",
      "weighted avg       0.71      0.70      0.70       416\n",
      "\n"
     ]
    }
   ],
   "source": [
    "lr = LogisticRegression(solver='liblinear', random_state=91)\n",
    "C_vals = [.0001,.001,.01,.05, .1,.12,.13,.135,.15,.16,.17,1,3,5,10,20,50,100]\n",
    "penalties = ['l1','l2']\n",
    "\n",
    "# Using GridSearch to find best penalty and C parameters\n",
    "gs = GridSearchCV(lr, {'penalty': penalties, 'C': C_vals}, verbose=False, cv=5)\n",
    "gs.fit(X_train1, y_train)\n",
    "print(gs.best_params_)\n",
    "y_pred_lr1 = gs.best_estimator_.predict(X_test1)\n",
    "y_predp_lr1 = gs.best_estimator_.predict_proba(X_test1)[:, 1]\n",
    "print(classification_report(y_test, y_pred_lr1, target_names=[\"Below Median\",\"Above Median\"]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.7043269230769231\n",
      "ROC_AUC: 0.7327260505759914\n",
      "Cross Val: 0.7137889302976638\n"
     ]
    }
   ],
   "source": [
    "acc_lr1 = accuracy_score(y_pred_lr1, y_test)\n",
    "print(\"Accuracy:\", acc_lr1)\n",
    "roc_lr1 = roc_auc_score(y_test, y_predp_lr1)\n",
    "print(\"ROC_AUC:\", roc_lr1)\n",
    "print(\"Cross Val:\", cross_val_score(gs.best_estimator_, X_train1, y_train, cv=5).mean())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Using Random Forest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "Below Median       0.76      0.65      0.70       219\n",
      "Above Median       0.67      0.77      0.72       197\n",
      "\n",
      "   micro avg       0.71      0.71      0.71       416\n",
      "   macro avg       0.71      0.71      0.71       416\n",
      "weighted avg       0.72      0.71      0.71       416\n",
      "\n"
     ]
    }
   ],
   "source": [
    "rf = RandomForestClassifier(random_state=91)\n",
    "rf.fit(X_train1, y_train)\n",
    "y_pred_rf1 = rf.predict(X_test1)\n",
    "y_predp_rf1 = rf.predict_proba(X_test1)[:,1]\n",
    "print(classification_report(y_test, y_pred_rf1, target_names=[\"Below Median\",\"Above Median\"]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.7091346153846154\n",
      "ROC_AUC: 0.7555107433419095\n",
      "Cross Val: 0.7168871101834084\n"
     ]
    }
   ],
   "source": [
    "acc_rf1 = accuracy_score(y_pred_rf1, y_test)\n",
    "print(\"Accuracy:\", acc_rf1)\n",
    "roc_rf1 = roc_auc_score(y_test, y_predp_rf1)\n",
    "print(\"ROC_AUC:\", roc_rf1)\n",
    "print(\"Cross Val:\", cross_val_score(rf, X_train1, y_train, cv=5).mean())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>importance</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Junior Executive</th>\n",
       "      <td>0.239000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Executive</th>\n",
       "      <td>0.171981</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Fresh/entry level</th>\n",
       "      <td>0.155085</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Professional</th>\n",
       "      <td>0.087981</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Senior Management</th>\n",
       "      <td>0.085051</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Non-executive</th>\n",
       "      <td>0.077965</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Senior Executive</th>\n",
       "      <td>0.064126</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Middle Management</th>\n",
       "      <td>0.061433</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Manager</th>\n",
       "      <td>0.057378</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                   importance\n",
       "Junior Executive     0.239000\n",
       "Executive            0.171981\n",
       "Fresh/entry level    0.155085\n",
       "Professional         0.087981\n",
       "Senior Management    0.085051\n",
       "Non-executive        0.077965\n",
       "Senior Executive     0.064126\n",
       "Middle Management    0.061433\n",
       "Manager              0.057378"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "features_importance = pd.DataFrame(rf.feature_importances_, index=X_train1.columns,\n",
    "                                  columns = [\"importance\"]).sort_values(\"importance\", ascending = False)\n",
    "features_importance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 40 candidates, totalling 120 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done  26 tasks      | elapsed:    7.4s\n",
      "[Parallel(n_jobs=-1)]: Done  97 out of 120 | elapsed:    7.6s remaining:    1.7s\n",
      "[Parallel(n_jobs=-1)]: Done 120 out of 120 | elapsed:    7.6s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'criterion': 'gini', 'max_depth': 3, 'n_estimators': 5}\n",
      "0.7086776859504132\n",
      "Score on test data: 0.7115384615384616\n"
     ]
    }
   ],
   "source": [
    "# Tuning of RandomForest model for better results\n",
    "params = {\n",
    "    'n_estimators' : range(1, 11, 2),\n",
    "    'criterion' : ['gini', 'entropy'],\n",
    "    'max_depth' : [1, 2, 3, None]\n",
    "}\n",
    "\n",
    "rf_gs = GridSearchCV(rf, param_grid=params, cv=3, n_jobs=-1, verbose=1)\n",
    "rf_gs.fit(X_train1, y_train)\n",
    "\n",
    "print(rf_gs.best_params_)\n",
    "print(rf_gs.best_score_)\n",
    "\n",
    "# Score test data with tuned parameters \n",
    "rfc2 = RandomForestClassifier(n_estimators=rf_gs.best_params_['n_estimators'], \\\n",
    "                              criterion=rf_gs.best_params_['criterion'], \\\n",
    "                              max_depth=rf_gs.best_params_['max_depth'], n_jobs=-1, random_state=29, verbose=1)\n",
    "rfc2.fit(X_train1, y_train)\n",
    "\n",
    "print('Score on test data: {}'.format(rfc2.score(X_test1, y_test)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "Below Median       0.77      0.65      0.70       219\n",
      "Above Median       0.67      0.78      0.72       197\n",
      "\n",
      "   micro avg       0.71      0.71      0.71       416\n",
      "   macro avg       0.72      0.72      0.71       416\n",
      "weighted avg       0.72      0.71      0.71       416\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(y_test, rfc2.predict(X_test1), labels=[0,1], target_names=['Below Median', 'Above Median']))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Using K Nearest Neighbours to classify"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "Below Median       0.62      0.47      0.54       219\n",
      "Above Median       0.54      0.68      0.60       197\n",
      "\n",
      "   micro avg       0.57      0.57      0.57       416\n",
      "   macro avg       0.58      0.58      0.57       416\n",
      "weighted avg       0.58      0.57      0.57       416\n",
      "\n"
     ]
    }
   ],
   "source": [
    "knn = KNeighborsClassifier(n_neighbors=2)\n",
    "knn.fit(X_train1, y_train)\n",
    "y_pred_kn1 = knn.predict(X_test1)\n",
    "y_predp_kn1 = knn.predict_proba(X_test1)[:,1]\n",
    "print(classification_report(y_test, y_pred_kn1, target_names=[\"Below Median\",\"Above Median\"]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.5697115384615384\n",
      "ROC_AUC: 0.5719931390955659\n",
      "Cross Val: 0.5598003339181695\n"
     ]
    }
   ],
   "source": [
    "acc_kn1 = accuracy_score(y_pred_kn1, y_test)\n",
    "print(\"Accuracy:\", acc_kn1)\n",
    "roc_kn1 = roc_auc_score(y_test, y_predp_kn1)\n",
    "print(\"ROC_AUC:\", roc_kn1)\n",
    "print(\"Cross Val:\", cross_val_score(knn, X_train1, y_train, cv=5).mean())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th>Summary:</th>\n",
       "      <th>LogisticReg</th>\n",
       "      <th>RandomForest</th>\n",
       "      <th>KNearest</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Accuracy Score</th>\n",
       "      <td>0.704327</td>\n",
       "      <td>0.709135</td>\n",
       "      <td>0.569712</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ROC AUC Score</th>\n",
       "      <td>0.732726</td>\n",
       "      <td>0.755511</td>\n",
       "      <td>0.571993</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "Summary:        LogisticReg  RandomForest  KNearest\n",
       "Accuracy Score     0.704327      0.709135  0.569712\n",
       "ROC AUC Score      0.732726      0.755511  0.571993"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.DataFrame([[acc_lr1, acc_rf1, acc_kn1],\n",
    "               [roc_lr1, roc_rf1, roc_kn1]],\n",
    "            columns=[\"LogisticReg\", \"RandomForest\", \"KNearest\"],\n",
    "            index=[\"Accuracy Score\", \"ROC AUC Score\"]).rename_axis('Summary:', axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Case 1 Conclusion:\n",
    "- Random Forest model gave the best accuracy score of 0.709, based on Job Levels as predictor variables.  Gridsearch for Random Forest barely improved the performance. \n",
    "- With the 1st simple model done, I'll explore if additional features such as Job Title can further improve the model's score."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#####  Use ADA-BOOST Classifier (random forest as base estimator) ..... to see if improves the best score of 0.709"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=-1)]: Using backend ThreadingBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n",
      "[Parallel(n_jobs=5)]: Using backend ThreadingBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done   2 out of   5 | elapsed:    0.0s remaining:    0.0s\n",
      "[Parallel(n_jobs=5)]: Done   5 out of   5 | elapsed:    0.0s finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.7091346153846154\n",
      "ROC_AUC: 0.7122708202952971\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "Below Median       0.76      0.65      0.70       219\n",
      "Above Median       0.67      0.77      0.72       197\n",
      "\n",
      "   micro avg       0.71      0.71      0.71       416\n",
      "   macro avg       0.71      0.71      0.71       416\n",
      "weighted avg       0.72      0.71      0.71       416\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Use Ada Boost Classifier with random forest (GV) as base estimator\n",
    "\n",
    "from sklearn.ensemble import AdaBoostClassifier\n",
    "abr=AdaBoostClassifier(n_estimators=50, base_estimator=rfc2,learning_rate=1.1)\n",
    "abr.fit(X_train1,y_train)\n",
    "y_pred_abr = abr.predict(X_test1)\n",
    "\n",
    "acc_rfc2 = accuracy_score(y_pred_abr, y_test)\n",
    "print(\"Accuracy:\", acc_rfc2)\n",
    "roc_rfc2 = roc_auc_score(y_test, y_pred_abr)\n",
    "print(\"ROC_AUC:\", roc_rfc2)\n",
    "print(classification_report(y_test, y_pred_abr, target_names=[\"Below Median\",\"Above Median\"]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#####  Use XG Boost Classifier  ..... to see if improves the best score of 0.709"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.7019230769230769\n",
      "ROC_AUC: 0.705166539183645\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "Below Median       0.75      0.64      0.69       219\n",
      "Above Median       0.66      0.77      0.71       197\n",
      "\n",
      "   micro avg       0.70      0.70      0.70       416\n",
      "   macro avg       0.71      0.71      0.70       416\n",
      "weighted avg       0.71      0.70      0.70       416\n",
      "\n"
     ]
    }
   ],
   "source": [
    "import xgboost as xgb\n",
    "xg_reg = xgb.XGBClassifier(objective ='reg:linear', \\\n",
    "                colsample_bytree = 0.3, learning_rate = 1,\\\n",
    "                max_depth = 5, alpha = 10, n_estimators = 10)\n",
    "xg_reg.fit(X_train1,y_train)\n",
    "y_preds_xg = xg_reg.predict(X_test1)\n",
    "acc_xg = accuracy_score(y_preds_xg, y_test)\n",
    "print(\"Accuracy:\", acc_xg)\n",
    "roc_xg = roc_auc_score(y_test, y_preds_xg)\n",
    "print(\"ROC_AUC:\", roc_xg)\n",
    "print(classification_report(y_test, y_preds_xg, target_names=[\"Below Median\",\"Above Median\"]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Both ADA-BOOST and XG_BOOST Classifiers did not improves the score any more. Hence we can conclude the optimal model is using Random Forest. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "-----"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Case 2:  Using `Job Levels` + `Job Title` to predict `Salary_Category`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Defining feature for Case 2\n",
    "features2 = df[\"Job Title\"]\n",
    "y = df[\"Salary_Category\"]\n",
    "\n",
    "features2_train, features2_test, y_train, y_test = train_test_split(features2, y, test_size=0.3, random_state=91)\n",
    " \n",
    "# Using CountVectorizer for processing Job Titles\n",
    "vect = CountVectorizer(lowercase=True,stop_words=stop, ngram_range=(1,1))\n",
    "features2_train = vect.fit_transform(features2_train)\n",
    "features2_test = vect.transform(features2_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# Concatenating previous Job Levels with Job Titles\n",
    "X_train2 = pd.concat([X_train1, pd.DataFrame(features2_train.toarray(), columns=vect.vocabulary_, index=X_train1.index)], axis=1)\n",
    "X_test2 = pd.concat([X_test1, pd.DataFrame(features2_test.toarray(), columns=vect.vocabulary_, index=X_test1.index)], axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'C': 1, 'penalty': 'l1'}\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "Below Median       0.75      0.71      0.73       219\n",
      "Above Median       0.70      0.74      0.72       197\n",
      "\n",
      "   micro avg       0.72      0.72      0.72       416\n",
      "   macro avg       0.72      0.72      0.72       416\n",
      "weighted avg       0.73      0.72      0.72       416\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Using Logistic Regression\n",
    "lr = LogisticRegression(solver='liblinear', random_state=91)\n",
    "C_vals = [.0001,.001,.01,.05, .1,.12,.13,.135,.15,.16,.17,1,3,5,10,20,50,100]\n",
    "penalties = ['l1','l2']\n",
    "\n",
    "# Using GridSearch to find best penalty and C parameters\n",
    "gs = GridSearchCV(lr, {'penalty': penalties, 'C': C_vals}, verbose=False, cv=5)\n",
    "gs.fit(X_train2, y_train)\n",
    "print(gs.best_params_)\n",
    "y_pred_lr2 = gs.best_estimator_.predict(X_test2)\n",
    "y_predp_lr2 = gs.best_estimator_.predict_proba(X_test2)[:, 1]\n",
    "print(classification_report(y_test, y_pred_lr2, target_names=[\"Below Median\",\"Above Median\"]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.7235576923076923\n",
      "ROC_AUC: 0.7975801404631111\n",
      "Cross Val: 0.7478582258508544\n"
     ]
    }
   ],
   "source": [
    "acc_lr2 = accuracy_score(y_pred_lr2, y_test)\n",
    "print(\"Accuracy:\", acc_lr2)\n",
    "roc_lr2 = roc_auc_score(y_test, y_predp_lr2)\n",
    "print(\"ROC_AUC:\", roc_lr2)\n",
    "print(\"Cross Val:\", cross_val_score(gs.best_estimator_, X_train2, y_train, cv=5).mean())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "Below Median       0.69      0.76      0.72       219\n",
      "Above Median       0.70      0.61      0.65       197\n",
      "\n",
      "   micro avg       0.69      0.69      0.69       416\n",
      "   macro avg       0.69      0.69      0.69       416\n",
      "weighted avg       0.69      0.69      0.69       416\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Using Random Forest\n",
    "rf = RandomForestClassifier(random_state=91)\n",
    "rf.fit(X_train2, y_train)\n",
    "y_pred_rf2 = rf.predict(X_test2)\n",
    "y_predp_rf2 = rf.predict_proba(X_test2)[:,1]\n",
    "print(classification_report(y_test, y_pred_rf2, target_names=[\"Below Median\",\"Above Median\"]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.6899038461538461\n",
      "ROC_AUC: 0.7638550865725611\n",
      "Cross Val: 0.7055520812360726\n"
     ]
    }
   ],
   "source": [
    "acc_rf2 = accuracy_score(y_pred_rf2, y_test)\n",
    "print(\"Accuracy:\", acc_rf2)\n",
    "roc_rf2 = roc_auc_score(y_test, y_predp_rf2)\n",
    "print(\"ROC_AUC:\", roc_rf2)\n",
    "print(\"Cross Val:\", cross_val_score(rf, X_train2, y_train, cv=5).mean())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "Below Median       0.61      0.88      0.72       219\n",
      "Above Median       0.74      0.38      0.50       197\n",
      "\n",
      "   micro avg       0.64      0.64      0.64       416\n",
      "   macro avg       0.67      0.63      0.61       416\n",
      "weighted avg       0.67      0.64      0.62       416\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Using KNearest Neighbours\n",
    "knn = KNeighborsClassifier(n_neighbors=2)\n",
    "knn.fit(X_train2, y_train)\n",
    "y_pred_kn2 = knn.predict(X_test2)\n",
    "y_predp_kn2 = knn.predict_proba(X_test2)[:,1]\n",
    "print(classification_report(y_test, y_pred_kn2, target_names=[\"Below Median\",\"Above Median\"]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.6418269230769231\n",
      "ROC_AUC: 0.6940755163062374\n",
      "Cross Val: 0.6496257459400454\n"
     ]
    }
   ],
   "source": [
    "acc_kn2 = accuracy_score(y_pred_kn2, y_test)\n",
    "print(\"Accuracy:\", acc_kn2)\n",
    "roc_kn2 = roc_auc_score(y_test, y_predp_kn2)\n",
    "print(\"ROC_AUC:\", roc_kn2)\n",
    "print(\"Cross Val:\", cross_val_score(knn, X_train2, y_train, cv=5).mean())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th>Summary:</th>\n",
       "      <th>LogisticReg</th>\n",
       "      <th>RandomForest</th>\n",
       "      <th>KNearest</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Accuracy Score</th>\n",
       "      <td>0.723558</td>\n",
       "      <td>0.689904</td>\n",
       "      <td>0.641827</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ROC AUC Score</th>\n",
       "      <td>0.797580</td>\n",
       "      <td>0.763855</td>\n",
       "      <td>0.694076</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "Summary:        LogisticReg  RandomForest  KNearest\n",
       "Accuracy Score     0.723558      0.689904  0.641827\n",
       "ROC AUC Score      0.797580      0.763855  0.694076"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.DataFrame([[acc_lr2, acc_rf2, acc_kn2],\n",
    "              [roc_lr2, roc_rf2, roc_kn2]],\n",
    "            columns=[\"LogisticReg\", \"RandomForest\", \"KNearest\"],\n",
    "            index=[\"Accuracy Score\", \"ROC AUC Score\"]).rename_axis('Summary:', axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Case 2 Conclusion:\n",
    "- Logistic regression model here has the best accuracy score of 0.7236. This model is higher than model 1 of Randon forest score of 0.709. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#####  Use ADA-BOOST Classifier (Logistic Regression as base estimator) ..... to see if improves the best score of 0.723"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.7115384615384616\n",
      "ROC_AUC: 0.7120042648865402\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "Below Median       0.74      0.70      0.72       219\n",
      "Above Median       0.69      0.72      0.70       197\n",
      "\n",
      "   micro avg       0.71      0.71      0.71       416\n",
      "   macro avg       0.71      0.71      0.71       416\n",
      "weighted avg       0.71      0.71      0.71       416\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Use Ada Boost Classifier with Logistic Regression as base estimator\n",
    "\n",
    "from sklearn.ensemble import AdaBoostClassifier\n",
    "abr=AdaBoostClassifier(n_estimators=50, base_estimator=lr,learning_rate=1.4)\n",
    "\n",
    "abr.fit(X_train2,y_train)\n",
    "y_pred_abr = abr.predict(X_test2)\n",
    "\n",
    "acc_lr3 = accuracy_score(y_pred_abr, y_test)\n",
    "print(\"Accuracy:\", acc_lr3)\n",
    "\n",
    "roc_lr3 = roc_auc_score(y_test, y_pred_abr)\n",
    "print(\"ROC_AUC:\", roc_lr3)\n",
    "\n",
    "print(classification_report(y_test, y_pred_abr, target_names=[\"Below Median\",\"Above Median\"]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Use XG Boost Classifier (with random forest as base estimator) ..... to see if improves the best score of 0.709"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.7115384615384616\n",
      "ROC_AUC: 0.7150638574044457\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "Below Median       0.77      0.65      0.70       219\n",
      "Above Median       0.67      0.78      0.72       197\n",
      "\n",
      "   micro avg       0.71      0.71      0.71       416\n",
      "   macro avg       0.72      0.72      0.71       416\n",
      "weighted avg       0.72      0.71      0.71       416\n",
      "\n"
     ]
    }
   ],
   "source": [
    "import xgboost as xgb\n",
    "xg_reg = xgb.XGBClassifier(objective ='reg:linear', \\\n",
    "                colsample_bytree = 0.3, learning_rate = 0.5,\\\n",
    "                max_depth = 5, alpha = 10, n_estimators = 10)\n",
    "\n",
    "xg_reg.fit(X_train2,y_train)\n",
    "y_preds_xg = xg_reg.predict(X_test2)\n",
    "\n",
    "acc_xg = accuracy_score(y_preds_xg, y_test)\n",
    "print(\"Accuracy:\", acc_xg)\n",
    "\n",
    "roc_xg = roc_auc_score(y_test, y_preds_xg)\n",
    "print(\"ROC_AUC:\", roc_xg)\n",
    "\n",
    "print(classification_report(y_test, y_preds_xg, target_names=[\"Below Median\",\"Above Median\"]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Both ADA-BOOST and XG_BOOST Classifiers did not improves the score any more. Hence we can conclude the optimal model is using Logistic regression with a score of 0.723."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "------"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Case 3 Consideration: Could `Job Title` alone better predict `Salary Category`?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Defining feature for Case 2\n",
    "features2 = df[\"Job Title\"]\n",
    "y = df[\"Salary_Category\"]\n",
    "\n",
    "features2_train, features2_test, y_train, y_test = train_test_split(features2, y, test_size=0.3, random_state=91)\n",
    "\n",
    "# Using CountVectorizer for processing Job Titles\n",
    "vect = CountVectorizer(lowercase=True,stop_words=stop, ngram_range=(1,1))\n",
    "features2_train = vect.fit_transform(features2_train)\n",
    "features2_test = vect.transform(features2_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'C': 0.13, 'penalty': 'l2'}\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "Below Median       0.68      0.76      0.72       219\n",
      "Above Median       0.70      0.61      0.65       197\n",
      "\n",
      "   micro avg       0.69      0.69      0.69       416\n",
      "   macro avg       0.69      0.69      0.69       416\n",
      "weighted avg       0.69      0.69      0.69       416\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Using Logistic Regression\n",
    "lr = LogisticRegression(solver='liblinear', random_state=91)\n",
    "C_vals = [.0001,.001,.01,.05, .1,.12,.13,.135,.15,.16,.17,1,3,5,10,20,50,100]\n",
    "penalties = ['l1','l2']\n",
    "\n",
    "# Using GridSearch to find best penalty and C parameters\n",
    "gs = GridSearchCV(lr, {'penalty': penalties, 'C': C_vals}, verbose=False, cv=5)\n",
    "gs.fit(features2_train, y_train)\n",
    "print(gs.best_params_)\n",
    "y_pred_lr3 = gs.best_estimator_.predict(features2_test)\n",
    "y_predp_lr3 = gs.best_estimator_.predict_proba(features2_test)[:, 1]\n",
    "print(classification_report(y_test, y_pred_lr3, target_names=[\"Below Median\",\"Above Median\"]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.6899038461538461\n",
      "ROC_AUC: 0.7447558120668475\n",
      "Cross Val: 0.7086928843106397\n"
     ]
    }
   ],
   "source": [
    "acc_lr3 = accuracy_score(y_pred_lr3, y_test)\n",
    "print(\"Accuracy:\", acc_lr3)\n",
    "roc_lr3 = roc_auc_score(y_test, y_predp_lr3)\n",
    "print(\"ROC_AUC:\", roc_lr3)\n",
    "print(\"Cross Val:\", cross_val_score(gs.best_estimator_, features2_train, y_train, cv=5).mean())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "Below Median       0.68      0.70      0.69       219\n",
      "Above Median       0.65      0.63      0.64       197\n",
      "\n",
      "   micro avg       0.67      0.67      0.67       416\n",
      "   macro avg       0.67      0.67      0.67       416\n",
      "weighted avg       0.67      0.67      0.67       416\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Using Random Forest\n",
    "rf = RandomForestClassifier(random_state=91)\n",
    "rf.fit(features2_train, y_train)\n",
    "y_pred_rf3 = rf.predict(features2_test)\n",
    "y_predp_rf3 = rf.predict_proba(features2_test)[:,1]\n",
    "print(classification_report(y_test, y_pred_rf3, target_names=[\"Below Median\",\"Above Median\"]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.6682692307692307\n",
      "ROC_AUC: 0.7259115036042927\n",
      "Cross Val: 0.6745332747127524\n"
     ]
    }
   ],
   "source": [
    "acc_rf3 = accuracy_score(y_pred_rf3, y_test)\n",
    "print(\"Accuracy:\", acc_rf3)\n",
    "roc_rf3 = roc_auc_score(y_test, y_predp_rf3)\n",
    "print(\"ROC_AUC:\", roc_rf3)\n",
    "print(\"Cross Val:\", cross_val_score(rf, features2_train, y_train, cv=5).mean())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "Below Median       0.59      0.84      0.69       219\n",
      "Above Median       0.67      0.35      0.45       197\n",
      "\n",
      "   micro avg       0.61      0.61      0.61       416\n",
      "   macro avg       0.63      0.59      0.57       416\n",
      "weighted avg       0.63      0.61      0.58       416\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Using KNearest Neighbours\n",
    "knn = KNeighborsClassifier(n_neighbors=2)\n",
    "knn.fit(features2_train, y_train)\n",
    "y_pred_kn3 = knn.predict(features2_test)\n",
    "y_predp_kn3 = knn.predict_proba(features2_test)[:,1]\n",
    "print(classification_report(y_test, y_pred_kn3, target_names=[\"Below Median\",\"Above Median\"]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.6081730769230769\n",
      "ROC_AUC: 0.6940755163062374\n",
      "Cross Val: 0.6311272545982229\n"
     ]
    }
   ],
   "source": [
    "acc_kn3 = accuracy_score(y_test, y_pred_kn3)\n",
    "print(\"Accuracy:\", acc_kn3)\n",
    "roc_kn3 = roc_auc_score(y_test, y_predp_kn3)\n",
    "print(\"ROC_AUC:\", roc_kn2)\n",
    "print(\"Cross Val:\", cross_val_score(knn, features2_train, y_train, cv=5).mean())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th>Summary:</th>\n",
       "      <th>LogisticReg</th>\n",
       "      <th>RandomForest</th>\n",
       "      <th>KNearest</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Accuracy Score</th>\n",
       "      <td>0.689904</td>\n",
       "      <td>0.668269</td>\n",
       "      <td>0.608173</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ROC AUC Score</th>\n",
       "      <td>0.744756</td>\n",
       "      <td>0.725912</td>\n",
       "      <td>0.681049</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "Summary:        LogisticReg  RandomForest  KNearest\n",
       "Accuracy Score     0.689904      0.668269  0.608173\n",
       "ROC AUC Score      0.744756      0.725912  0.681049"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.DataFrame([[acc_lr3, acc_rf3, acc_kn3],\n",
    "              [roc_lr3, roc_rf3, roc_kn3]],\n",
    "            columns=[\"LogisticReg\", \"RandomForest\", \"KNearest\"],\n",
    "            index=[\"Accuracy Score\", \"ROC AUC Score\"]).rename_axis('Summary:', axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Case 3 Conclusion:\n",
    "- Logistic regression model here provided the highest score of 0.6899 as compared to other models.\n",
    "- `Job Title` itself is an important feature to predict Salary"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "------"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Case 4: Using `Job Levels` + `Job Title` + `Job Industry` to predict `Salary_Category`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "X4 = df[list2]\n",
    "y4 = df[\"Salary_Category\"]\n",
    "X_train4, X_test4, y_train, y_test = train_test_split(X4, y4, test_size=0.3, random_state=91)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Concatenating previous Job Levels with Job Titles\n",
    "X_train4 = pd.concat([X_train4, X_train2], axis=1)\n",
    "X_test4 = pd.concat([X_test4, X_test2], axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'C': 1, 'penalty': 'l1'}\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "Below Median       0.74      0.72      0.73       219\n",
      "Above Median       0.70      0.73      0.71       197\n",
      "\n",
      "   micro avg       0.72      0.72      0.72       416\n",
      "   macro avg       0.72      0.72      0.72       416\n",
      "weighted avg       0.72      0.72      0.72       416\n",
      "\n"
     ]
    }
   ],
   "source": [
    "lr = LogisticRegression(solver='liblinear', random_state=91)\n",
    "C_vals = [.0001,.001,.01,.05, .1,.12,.13,.135,.15,.16,.17,1,3,5,10,20,50,100]\n",
    "penalties = ['l1','l2']\n",
    "\n",
    "# Using GridSearch to find best penalty and C parameters\n",
    "gs = GridSearchCV(lr, {'penalty': penalties, 'C': C_vals}, verbose=False, cv=5)\n",
    "gs.fit(X_train4, y_train)\n",
    "print(gs.best_params_)\n",
    "y_pred_lr4 = gs.best_estimator_.predict(X_test4)\n",
    "y_predp_lr4 = gs.best_estimator_.predict_proba(X_test4)[:, 1]\n",
    "print(classification_report(y_test, y_pred_lr4, target_names=[\"Below Median\",\"Above Median\"]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.7211538461538461\n",
      "ROC_AUC: 0.8102241383306679\n",
      "Cross Val: 0.7468803852522671\n"
     ]
    }
   ],
   "source": [
    "acc_lr4 = accuracy_score(y_pred_lr4, y_test)\n",
    "print(\"Accuracy:\", acc_lr4)\n",
    "roc_lr4 = roc_auc_score(y_test, y_predp_lr4)\n",
    "print(\"ROC_AUC:\", roc_lr4)\n",
    "print(\"Cross Val:\", cross_val_score(gs.best_estimator_, X_train4, y_train, cv=5).mean())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "Below Median       0.73      0.69      0.71       219\n",
      "Above Median       0.68      0.72      0.70       197\n",
      "\n",
      "   micro avg       0.70      0.70      0.70       416\n",
      "   macro avg       0.70      0.71      0.70       416\n",
      "weighted avg       0.71      0.70      0.70       416\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Using Random Forest\n",
    "rf = RandomForestClassifier(random_state=91)\n",
    "rf.fit(X_train4, y_train)\n",
    "y_pred_rf4 = rf.predict(X_test4)\n",
    "y_predp_rf4 = rf.predict_proba(X_test4)[:,1]\n",
    "print(classification_report(y_test, y_pred_rf4, target_names=[\"Below Median\",\"Above Median\"]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.7043269230769231\n",
      "ROC_AUC: 0.7724659852119695\n",
      "Cross Val: 0.7251878940282337\n"
     ]
    }
   ],
   "source": [
    "acc_rf4 = accuracy_score(y_pred_rf4, y_test)\n",
    "print(\"Accuracy:\", acc_rf4)\n",
    "roc_rf4 = roc_auc_score(y_test, y_predp_rf4)\n",
    "print(\"ROC_AUC:\", roc_rf4)\n",
    "print(\"Cross Val:\", cross_val_score(rf, X_train4, y_train, cv=5).mean())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "Below Median       0.67      0.74      0.70       219\n",
      "Above Median       0.67      0.58      0.63       197\n",
      "\n",
      "   micro avg       0.67      0.67      0.67       416\n",
      "   macro avg       0.67      0.66      0.66       416\n",
      "weighted avg       0.67      0.67      0.67       416\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Using KNearest Neighbours\n",
    "knn = KNeighborsClassifier(n_neighbors=4)\n",
    "knn.fit(X_train4, y_train)\n",
    "y_pred_kn4 = knn.predict(X_test4)\n",
    "y_predp_kn4 = knn.predict_proba(X_test4)[:,1]\n",
    "print(classification_report(y_test, y_pred_kn4, target_names=[\"Below Median\",\"Above Median\"]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.6682692307692307\n",
      "ROC_AUC: 0.7071830887977193\n",
      "Cross Val: 0.662140500384184\n"
     ]
    }
   ],
   "source": [
    "acc_kn4 = accuracy_score(y_pred_kn4, y_test)\n",
    "print(\"Accuracy:\", acc_kn4)\n",
    "roc_kn4 = roc_auc_score(y_test, y_predp_kn4)\n",
    "print(\"ROC_AUC:\", roc_kn4)\n",
    "print(\"Cross Val:\", cross_val_score(knn, X_train4, y_train, cv=5).mean())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th>Summary:</th>\n",
       "      <th>LogisticReg</th>\n",
       "      <th>RandomForest</th>\n",
       "      <th>KNearest</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Accuracy Score</th>\n",
       "      <td>0.721154</td>\n",
       "      <td>0.704327</td>\n",
       "      <td>0.668269</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ROC AUC Score</th>\n",
       "      <td>0.810224</td>\n",
       "      <td>0.772466</td>\n",
       "      <td>0.707183</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "Summary:        LogisticReg  RandomForest  KNearest\n",
       "Accuracy Score     0.721154      0.704327  0.668269\n",
       "ROC AUC Score      0.810224      0.772466  0.707183"
      ]
     },
     "execution_count": 67,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.DataFrame([[acc_lr4, acc_rf4, acc_kn4],\n",
    "              [roc_lr4, roc_rf4, roc_kn4]],\n",
    "            columns=[\"LogisticReg\", \"RandomForest\", \"KNearest\"],\n",
    "            index=[\"Accuracy Score\", \"ROC AUC Score\"]).rename_axis('Summary:', axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Conclusion for Case 4:\n",
    "- Accuracy score for all 3 models improved as compared to Case 2 where I did not add `Job Industry` in it.\n",
    "- Logistic Regression has the best overall accuracy score of 0.7211 with the best ROC/AUC score of 0.81 by far\n",
    "- In conclusion, we can conclude that 3 predictors \"Job Levels\", \"Job Title\", \"Job Industry\" is a much better model for predicting salary. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "------"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### QUESTION 2: Factors that distinguish job category\n",
    "\n",
    "Using the job postings you scraped for part 1 (or potentially new job postings from a second round of scraping), identify features in the data related to job postings that can distinguish job titles from each other. There are a variety of interesting ways you can frame the target variable, for example:\n",
    "- What components of a job posting distinguish data scientists from other data jobs?\n",
    "- What features are important for distinguishing junior vs. senior positions?\n",
    "- Do the requirements for titles vary significantly with industry (e.g. healthcare vs. government)?\n",
    "\n",
    "You may end up making multiple classification models to tackle different questions. Be sure to clearly explain your hypotheses and framing, any feature engineering, and what your target variables are. The type of classification model you choose is up to you. Be sure to interpret your results and evaluate your models' performance.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Defining Classification Problem (if Job Title contains Scientist)\n",
    "df[\"is_datascientist\"] = 0\n",
    "df.loc[(df[\"Job Title\"].str.lower().str.contains(\"scientist\")), \"is_datascientist\"] = 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Distribution Plot of Class 0 & 1\n",
    "df[\"is_datascientist\"].value_counts().plot.bar()  \n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Baseline: 0.9342485549132948\n"
     ]
    }
   ],
   "source": [
    "# Identifying model baseline\n",
    "baseline6 = len(df[df[\"is_datascientist\"] == 0]) / len(df[\"is_datascientist\"])\n",
    "print(\"Baseline:\", baseline6)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Using Job Description to predict whether Job is Data Scientist or not\n",
    "X6 = df['Job Description']\n",
    "y6 = df['is_datascientist']\n",
    "desc1_train, desc1_test, y_train, y_test = train_test_split(X6, y6, test_size=0.3, random_state=91)\n",
    "\n",
    "cvz = TfidfVectorizer(stop_words=stop, ngram_range=(1, 1), max_features=200)\n",
    "stop += [\"requirements\", \"work\"]\n",
    "desc1_train = cvz.fit_transform(desc1_train, y_train)\n",
    "desc1_test = cvz.transform(desc1_test)\n",
    "\n",
    "desc1_traindf = pd.DataFrame(desc1_train.toarray(), columns=cvz.get_feature_names(), index=X_train1.index)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAbUAAAEICAYAAADY/mp2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvOIA7rQAAIABJREFUeJzt3Xm8XfO9//HXO4SERGIeKoQgqSohh5qFqmsqNV1aWkqlbluqvdqr5ZJWtYbezpQYyv1VVY1V2qAkEjGekEhiLOGaYwohpiSf3x/f72HZ9jlnJznn7L2X9/Px2I+z13d913d91tpr78/+ftc6eykiMDMzK4Ne9Q7AzMysqzipmZlZaTipmZlZaTipmZlZaTipmZlZaTipmZlZaTipWdOTdJqk8xdhucckbdUdMdmikXSLpAO7qe2LJH2/O9ruYJ1HSfpnT66zK0g6QtLfFmP5Ht/XbZzUMklvFB4LJL1VmD64G9a3h6TbJM2RNCu/mXfr6vVY+yJiSETcsajL52S6WmF6mKSQdGVFvSskHb84sVa0d1En8/8s6Z18bM2RdL+kUyT1W4h1PC9p28WIcRNJN0t6NT/ukbRzZ8tFxE4Rcdmirrew/o8kk4g4LCLOqGHZDrc9v87zFjfGDtY9N79ur0qaKOlrktQd62tPRFwQEZ+vpe7i7Ovu4KSWRUS/tgfwf8DnC2WXdOW6cpL8E3Ae8AlgdeBUYO+uXE/ZSOolqe7HrKSTCj283pJGS9osTy8AdpQ0oovXuZykcyQtn6fXlXR2B4ucEhH9gZWBI4EdgYmS+nRlXO3E2gu4HvgrsAqwGvCfwJvdve6S2CW/dusAvwROAjp6rbuUpCV7al3dIiL8qHgATwA7V5T1Bc4CngOeBs4Eeud5uwL/An4EvAI8DhzQTttLAs8DR3ew/iVyW/8HvABcCPTP84YB84AjgGeAl4HDga2A6cBs4BeFto4CbgF+B7wGPAq0AKPy8i8ABxXqr0BKuC8CM4HvAyq0dTPwm7yexyr3U6Gd/wAuL0w/BfxvYXoWMCw/3wG4N8d3J7B5od6dwI+Bu4C3gTWB9YBJwBzgH8C5wPm5/rLAn/PrMDsvt3w7MT4PbJufnwZcAlya270fGN7Ocv2BM4BHgLHAnhWvzQnAPwr1rwCOL0x/M++7l4GrgFVzeR8gSEnoMeBV4JeF5YYDf8nH1x+AQe3E92fgxIqy5fNr+rVCrOPzfnoRuLhwjF1OSs5zgTeAY0jH7ZX5eJkNjAOGtrP+NfN29O3gGD8g7+PXScfkZwuv9yGFel8HHs5xXg98orN9BWyaj5V5Of7nK/cLKdGOzdvyMnBLe9teJfZhwLyFeM/cQjpGXwceALbvYL+8f0wWyrbLMa1f+Cz6Fek99TzwW2DpjrYrzxtM+qLxUn78T0WMZ+X9eGIu+2fFvv4W6bPxRdKXcNWyrxfzmB8G3Eb6bHiRwmdIu/uw1g/6j9OD6kntDGAisBKwKnAPcEKet2t+UX8GLAXsnN8U61Rpe3h+EVfvYP3fAB4E1gaWA64Dziu8yAH8Glga2Iv0DfhKYEVgrXxgfKZwwL4HfIn0wXQm8CTpG+BSeflXgT65/l9Ib+x+pOQxEzi4oq2vkBLvd4An2tmGDYFZ+fm6uZ0nCvNeyM9XIb3Z/z3Hd1g+eAfk+XeSPsSHAr1znXsL+/qzeV+3JbVvk5JI31x3c2DZzj5ASEltLvC5vG2/BMa3s1z/XP8RUlLdrfDazMv7blah7feTGrB7Xu/GpDf0GOCmijf4Vfl1X4f04TSycOy0JbULyB/wVeL7SFIrvLYXF2LdKe/D1fJ+Pq3avsnTSwKH5m3rA/weuLOd9S9Jeg9dTRp9WKVi/nakY25H0mjRWsAGhdf7kPz8INL7YIP82v8EGFfjvnr/Q7nafsmv769zrEtRSDSV215l+yqTWmfvmXmk93Rv0nvnFWC5zo7JivJZwFfz83PyMTUQGADcAJzc0XbldT9IOm6XIb0/tq6I8UjSsd+X6knthrzOdUjH4CE17uvFOeavBo4jJdC+wDadfn4v7Af+x+FB9aT2DLBTYXpv4KH8fFfSt5U+hfnXAt+r0vZnSd+6enWw/knA4YXpTUgfuOKDpLZiYf6bwN6F6euBowoH3LTCvM3z8gMqlh9GSpLzgXUL874NjC20Nb0wb4Xc1sB2tmMWKYEdRurd3U/6tvgfwF9ynSOBCRXL3UfuPZI+5H5YmLdBlX19FR8ktW8AtwIb1fA6Vya16wrzNgNmt7PcyaSe8WmkD+Qfkb6xvv9hB3wXuDU/Lya1S4AfF9oamI+H1fjgDd5ScRwdS3rDn0PqcV0EDAHOaie+9pLar4C/tbPMQcAd1fZNO/VXy3H3aWf+2jnemfmYupn8JY/UK/xZO8sVk9o4cnLI071JX6pW7WhfFY7Vjj5ozyAlonWrxFBzUqO298zMiuXvp/2RnPaS2hTSEO6SwLsUvtCQvhw82NF25TrPUOVzJ8f4SJWyyqQ2sjD/u8D1Ne7rRTrm8/O/kEaZ2u0EVD7qfn6iGeSTtKuRejhtniSdD2vzYkS8XTF/jSrNvUxKTqt2sMo1qqyrLymJAMyPiJcL898iDQsVp4sXBVTOeyciXqtSfzXSN+f/q1h3cTufLzyfm/+2dwHCBGAksD0p0YwnDTXukKfho9tabZ1PFZ6vQfV93eaC3PYVkp6W9FNJS7QTX6XKbau6XRHxo/jgApN3I+LkiLivotrZwBBJn6so/9D2RsRsUk+1o33cLyJej4ijIuLVvNxjEfHNGrerzSdIvQQkrSHpcknPSHodOJ80ClGVpCUl/VzS47n+Q6TjeMVq9SPiyRzvOqSeOqRhdIBBpKGmzqwNnCNptqTZpB78PNLwZpuaXrMqTgWeBcZJ+pek79a4XKVa3jNPVyzT3mdDR9peuzVIyX1GYb9cQxrxgPa3axApuS5op/2n2ilvr87CbMMiHfP5+XdIPcv78gVPh3S2Mie1GkT6yvA86U3WZi3SN582K1WchF+LdHBVmk5KMvt1sMpnq6zrLfIHUjd6nvQNaq2KdT9TvXqnbiUlte1ICe5WUkJrS3Lw0W2tts4oPH+O6vs6VYx4JyJOiohheT0HkHohXS4ijo+I59uZ9zZpuOwnFbM+tL2SBpB6YTXv44g4bGFjlTSQ9FpMzEVnknroG0XEcsDXSEnq/dVUNPFVYBfSN/4BpN4KFcu0F++TpOHKjXLRU6SeZmeeAg6LiIGFR9+ImFzDspXxV8b0WkR8OyLWJr0XT5S0TS3LVqjlPbMmH9beZ0NV+UrMFUnnlp4jJfYhhX0yICJW7GS7ngIGd3ChVS3bPKidbehs2UU+5iPimYg4nHQx3THAhZLW6mgZJ7XaXQqcLGlFSauQLgb4Y2F+b+C/JS0laSfSuZkrKxuJiHmkMeKfSPqypP75qr4dClezXQocJ2ktSf1JH4x/ysm120TEO6Qx7J9KWlbSENJQyh87XrJdt5KGZt+NiBfz9H6ksf4Zuc61wKaS9s+9ga+Q3jBj22nzEVIvoW1f75jXAYCknSVtmN+8r5M+AOYvYvyL6wLS1Yc7FsouBY6UtFFOzKeTTuZXTY6LS1IfSVuQLhB4lg9ey/6kE/uv5w+Jyp7KC3zQw2qr/zZppGFZPpqsi+tcNV8huq6SVUhD0HfmKucDX5e0fT72B0naoEpT55A+lIfmdpeX1NGXwcr4B0nq3U6Me0laJ4/CvEY6RuYXll232nKVanzPDMqXvS+ZexprATd21rakAZK+kNs6PyIejYj3SD3eX0taKe/fQW0jAh1s122kC6BOkbSMpL6Stq5lGwv+K8c0mHTRSNu/XnS4r1mMY17SgZLWyJ99s3Nxh/9O4aRWu5NIVy7NII1vTyKNX7d5grSznycddF+NiMerNRQRfwQOIY1FP5eXOZn0wQPpW+1VwO2kYZpX+OiHTnf5ev77JOmKqPNJY+KLYhrpHMgEgIh4ifSNcWJbgo6IF0gXq5xA+sD8FulqwtnVGszLHUhKFK+QrjQrfoB8grQf55B6xX8njcv3uPwBNJoPho2JiOtIF7lcS0oyqwFf7obV/7ekOaSr3C4kHa/bFYZtTwK2JX3wXc1Hv4CdCpyah7i+RUrQL5KO1WmkD8n2vA2sTxpungNMJV0Y8jWAiJhIOvbPzuu/mY/2ZoiIS0nnU67KQ55TSF8WazGW9J6cJaly+A/gk4X4JgA/j4i2pFu57dUUv2B29p6ZQDrn+grpON+nYvi/0o2S3sjtfY90vBxVmH8s6dhpJe2/saQLVNrdrnws7k46P/80abh03w5iqOZ60mvZSjpv1/a+63BfL+YxvxUwOe+Py4FREdFhL1fd/OX/Y0HSrsDvImK9TiubWVPLPd9rI2K1TiuXQO5dvUX6F5JqXxAaintqZmY1ysPaB5B6K9aAmvs/x83MetYs0imBQ+sdiFXn4UczMysNDz+amVlpePixC6y00koxePDgeodhZtZUJk+e/FJErNyVbTqpdYHBgwfT2urzxmZmC0NS5a8JLTYPP5qZWWk4qZmZWWk4qZmZWWk4qZmZWWn4QpEuMHXOXFYbN6XeYZiZ9ajndxxe7xA+wj01MzMrjbomNUmDJU1fzDbWkHRFV8VkZmbNq+mHH/NtCPavdxxmZlZ/jTD8uKSki/Otuq/IN7B7QtJKAJJaJI3Pz3eQNCU/7ss32Hy/tyfpMElXSRor6VFJ79/vTNIuku6QdK/Sbez75fLTJD2Q1//zXHaApOmSpkqa0ON7xMzMFkkj9NSGAkdExCRJFwLf6KDuccA3c91+pJsRVhpOuiHfO8DDkn5LuhfQicDOEfGmpP8Cvivpd8A+wLCICKVb3kO6geK/RcQzhbIPkTQKGAXQa9XVF3abzcysGzRCT+2piJiUn/+RdDfe9kwCfiHpGGBgRFS7rffNEfFavsPvA8DawJbAhsAkSVNIt41YG3idlBjPl7QvMLewnoskHQksUS2QiBgTES0R0dJrQNW8Z2ZmPawRklrlvW8CmMcHsfV5f0bEaaRbwvcF7pQ0rEp77xSezyf1RgXcFBHD82PDiDgiJ8UtSLey/wLptuRExFGknt0gYIqkFRdzG83MrAc0QlJbS9JW+fkXgduAJ4ARuWy/toqShkTEtIg4nXTn2WpJrZo7gW0krZfbWUbSBnkIc0BE/B04ljR02baeuyLiJOAlUnIzM7MG1wjn1B4EDpV0LvAo8HvgbuACST8E7irUPVbSjqQe2APAP4BOT2hFxIuSDgMulbR0Lj4RmAP8VVIfUm/uO3nemZLWz2U3A1MXbxPNzKwn+M7XXaD30A1jxXP+VO8wzMx61OL+ooikyRHR0kXhAI3RU2t6m/RfhtYG/LkYM7OPm0Y4p2ZmZtYlnNTMzKw0nNTMzKw0nNTMzKw0nNTMzKw0nNTMzKw0nNTMzKw0nNTMzKw0nNTMzKw0nNTMzKw0/DNZXWDOnGncfMuQeodhZg3kszs9Vu8QPpbcUzMzs9Jo+qQmaaCkb9Q7DjMzq7+mT2rAQMBJzczMSpHUTgOGSJoi6UxJ35N0j6T7Jf2orZKkayRNljRD0qhC+RuSTs/z/ilpC0njJT0uaa+6bJGZmS2SMiS144HHImI4cBOwPrAFMBwYIWn7XO/wiBgBtADHSFoxly8LjM/z5gA/AT4H7AP8uL2VSholqVVS6+zZC7pju8zMbCGV7erHXfLjvjzdj5TkJpAS2T65fFAufxl4Fxiby6cB70TEe5KmAYPbW1FEjAHGAAwdurRvH25m1gDKltQE/Cwizv1QoTQS2BnYKiLmShoP9Mmz34uItqS0AHgHICIWSCrb/jEzK7UyDD/OAfrn5zcAh0vqByDpE5JWAQYAr+aENgzYsj6hmplZd2r6nkhEvCxpkqTpwD+APwF3SAJ4AziENLx4lKT7gYeBO+sVr5mZdR99MPJmi6qlpSVaW1vrHYaZWVORNDkiWrqyzTIMP5qZmQFOamZmViJOamZmVhpOamZmVhpOamZmVhpOamZmVhpOamZmVhpOamZmVhpOamZmVhpOamZmVhpN/9uPjeDZZ59l9OjR9Q7DzBqIPxPqwz01MzMrDSc1MzMrDSe1dkg6VtIy9Y7DzMxq56RWhaQlgGMBJzUzsybScElN0rKSrpc0VdJ0SQdKekLSSnl+i6Tx+floSf9P0i2SHpV0ZC4fKWmCpKslPSDpHEm98rwvSpqW2z69sN43JP1Y0l3ACcAawDhJ43p6H5iZ2aJpxKsfdwWejYg9ACQNAE7voP7GwJbAssB9kq7P5VsAGwJPku58va+k23NbI4BXgRslfSEirsnLT4+Ik/J6Dwd2jIiXqq1U0ihgFMCAAQMWY3PNzKyrNFxPDZgG7CzpdEnbRcRrndT/a0S8lZPPOFIyA7g7Ih6PiPnApcC2wObA+Ih4MSLmAZcA2+f684Eraw0yIsZEREtEtCyzjEcpzcwaQcP11CLiEUkjgN2Bn0m6EZjHBwm4T+Ui7UxXK1cHq347J0AzM2tSDddTk7QGMDci/gj8HNgMeII0ZAiwX8Uie0vqI2lFYCRwTy7fQtI6+VzagcBtwF3ADpJWyheDfBG4tZ1Q5gD9u2arzMysJzRcTw34NHCmpAXAe8B/AH2BCyT9kJSYiu4GrgfWAk6JiGclbQDcAZyW25sAXB0RCyT9gDRMKeDvEfHXduIYA/xD0nMRsWPXbqKZmXUHRVSO0jUPSaOBNyLi5xXlI4HjImLPnoijpaUlWltbe2JVZmalIWlyRLR0ZZsNN/xoZma2qBpx+LFmETG6nfLxwPiejMXMzOrPPTUzMysNJzUzMysNJzUzMysNJzUzMysNJzUzMysNJzUzMysNJzUzMysNJzUzMyuNpv7n60bx7jNv8PTxE+sdhpnVwZqnbVfvEKzAPTUzMysNJ7V2SBos6Uv1jsPMzGrnpNa+wYCTmplZE+k0qeUey0OSzpc0XdIlknaWNEnSo5K2yI/bJd2X/w7Nyx4m6SpJY3PdMwrt/l5Sq6QZkn5UKN89r+82Sb+RdF0uX1bShZLuyevZu7COayT9TdJMSd+S9N1c505JK+R6Q3IckyVNlDQsl1+U13O7pMcl7Z9DOQ3YTtIUSd/pqh1uZmbdp9ae2nrAr4GNgWGkHsy2wHHAD4GHgO0jYlPgJOCnhWWHk+48/WngQEmDcvkJ+T46G5PuRr2xpD7AucBuEbEtsHKhnROAWyJic2BH0o1El83zNsoxbQGcSrpz9qakG4V+JdcZAxwdESNy3GcX2l49b8+epGQGcDwwMSKGR8QvK3eIpFE5Kbe+Mnd253vQzMy6Xa1XP86MiGkAkmYAN0dESJpGGqYbAFwsaX0ggN6FZW+OiNfysg8AawNPAf8uaVSOYXVgQ1KSfTwiZuZlLwVG5ee7AHtJOi5P9yHd7RpgXETMAeZIeg34Wy6fBmwsqR+wNXC5pLa4li7EeE1ELAAekLRqLTskIsaQEiUbrz6see+0amZWIrUmtXcKzxcUphfkNk4hJZZ9JA3mw/cyKy47H1hS0jqk3tLmEfGqpItISUq0T8B+EfHwhwqlz9QQXy9gdkQMr2H7OorBzMwaWFddKDIAeCY/P6yG+ssBbwKv5Z7Rbrn8IWDdnBghDVu2uQE4WrmrJWnTWoOLiNeBmZIOyMtK0iadLDYH6F/rOszMrP66KqmdAfxM0iRgic4qR8RU4D5gBnAhMCmXvwV8Axgr6TbgBeC1vNgppGHN+yVNz9ML42DgCElT83r37qT+/cA8SVN9oYiZWXNQRGOdDpLULyLeyD2ys4BHq12o0Ug2Xn1Y/P3Q8+odhpnVgX9RZNFJmpwvGOwyjfgzWUdKOhRYitSbO7fO8XRqqU/084FtZtYAGi6p5V5ZQ/fMzMysMfkXRczMrDSc1MzMrDSc1MzMrDSc1MzMrDSc1MzMrDSc1MzMrDSc1MzMrDSc1MzMrDSc1MzMrDQa7hdFmtELj/+L/zlwz3qHYWYL6T8vu67eIVgXc0/NzMxKo+5JTdLowt2sq83/gqQNa2jnIkn7d210ZmbWTOqe1GrwBaDTpGZmZlaXpCbpBEkPS/onMDSXHSnpnnxTzislLSNpa2Av4ExJUyQNqVav0PTOkiZKekTSnrndwbns3vzYOpevLmlCbne6pO1y+S6S7sh1L5fUr2f3jpmZLaoeT2qSRgAHAZsC+wKb51lXRcTmEbEJ8CBwRETcDlwLfC8ihkfEY9XqFZofDOwA7AGcI6kPMAv4XERsBhwI/CbX/RJwQ0QMBzYBpkhaCTgR2DnXbwW+2852jJLUKqn1zXfe7YI9Y2Zmi6seVz9uB1wdEXMBJF2byzeS9BNgINAPuKGd5Tuq95eIWAA8KulxYBgwE/idpOHAfGCDXPce4EJJvYFrImKKpB1IQ52T0o23WQq4o1oQETEGGAMwaIWBjXX7cDOzj6l6XdJfLQlcBHwhIqZKOgwY2c6yHdWrbDeA7wAvkHpjvYC3ASJigqTtSb26/yfpTOBV4KaI+OJCb5GZmdVdPc6pTQD2kdRXUn/g87m8P/Bc7jkdXKg/J8+jk3oAB0jqJWkIsC7wMDAAeC734L4MLAEgaW1gVkScB1wAbAbcCWwjab1cZxlJG2BmZk2hx3tqEXGvpMuAKcCTwMQ867+Bu3LZND5IZH8GzpN0DLB/B/UgJbFbgVWBoyLibUlnA1dKOgAYB7yZ644EvifpPeAN4CsR8WLu/V0qaelc70Tgka7bA2Zm1l0U4dNBi2vQCgPj2M9tW+8wzGwh+RdF6kvS5Iho6co2/TNZXWDVddfzm8PMrAE0wz9fm5mZ1cRJzczMSsNJzczMSsNJzczMSsNJzczMSsNJzczMSsNJzczMSsNJzczMSsNJzczMSsNJzczMSsM/k9UFZj05h7OOuqXeYZg1vW+es1O9Q7Am556amZmVxiIlNUmjJR3XVUF0dXtdQdJwSbvXOw4zM6ude2rtGw44qZmZNZGak5qkEyQ9LOmfwNBcNkTSWEmTJU2UNEzSAElPSOqV6ywj6SlJvavVr7Ke4ZLulHS/pKslLZ/Lx0v6laTbJU2XtEUuHy3pYkk35vXuK+kMSdPyunrneiMk3ZrXfYOk1Qvtni7pbkmPSNpO0lLAj4EDJU2RdOBi7mczM+sBNSU1SSOAg4BNgX2BzfOsMcDRETECOA44OyJeA6YCO+Q6nwduiIj3qtWvsrr/Bf4rIjYm3dn65MK8ZSNia+AbwIWF8iHAHsDewB+BcRHxaeAtYI+c2H4L7J/XfSFwamH5JSNiC+BY4OSIeBc4CbgsIoZHxGVV9skoSa2SWt94e3aH+8/MzHpGrVc/bgdcHRFzASRdC/QBtgYul9RWb+n89zLgQGAcKRmeLalfB/XJ7Q4ABkbErbnoYuDyQpVLASJigqTlJA3M5f+IiPckTQOWAMbm8mnAYFLPciPgprzuJYDnCu1elf9OzvU7FRFjSEmatVYe6tuHm5k1gIW5pL/yg7sXMDsihlepey3wM0krACOAW4BlO6i/qDG0Tb8DEBELJL0XEW3lC0jbKGBGRGzVTrvv5L/z8b85mJk1rVrPqU0A9pHUV1J/0pDiXGCmpAMAlGwCEBFvAHcDvwaui4j5EfF6e/Xb5KHLVyVtl4u+DNxaqHJgXnZb4LVcvxYPAytL2iov31vSpzpZZg7Qv8b2zcysAdSU1CLiXtKQ4hTgSmBinnUwcISkqcAM0jmtNpcBh+S/1FC/zaHAmZLuJ12B+OPCvFcl3Q6cAxxRS+w5/neB/YHT87qnkIZCOzIO2NAXipiZNQ99MFLX2CSNB46LiNZ6x1KppaUlWlsbLiwzs4YmaXJEtHRlm/4/NTMzK42muSgiIkbWOwYzM2ts7qmZmVlpOKmZmVlpOKmZmVlpOKmZmVlpOKmZmVlpOKmZmVlpOKmZmVlpOKmZmVlpNM0/Xzeyt6fP4MFhn6x3GGZN75MPPVjvEKzJuadmZmal4aRmZmal0ZRJTdJoScct5DItkn7TXTGZmVn9fWzOqeVb1vj+MGZmJdY0PTVJJ0h6WNI/gaG5bIiksZImS5ooaVguP0DSdElTJU3IZSMlXZefryzpJkn3SjpX0pOSVpI0WNKDks6TNEPSjZL61m2jzcxsoTRFUpM0AjgI2BTYF9g8zxoDHB0RI4DjgLNz+UnAv0XEJsBeVZo8GbglIjYDrgbWKsxbHzgrIj4FzAb2ayemUZJaJbW+Mn/eYm2fmZl1jWYZftwOuDoi5gJIuhboA2wNXC6prd7S+e8k4CJJfwGuqtLetsA+ABExVtKrhXkzI2JKfj4ZGFwtoIgYQ0qqbNSnb3PcPtzMrOSaJakBVCaOXsDsiBj+kYoRR0n6DLAHMEVSZR1VLlPwTuH5fMDDj2ZmTaIphh+BCcA+kvpK6g98HpgLzJR0AICSTfLzIRFxV0ScBLwEDKpo7zbg33PdXYDle2g7zMysGzVFUouIe4HLgCnAlcDEPOtg4AhJU4EZwN65/ExJ0yRNJyXEqRVN/gjYRdK9wG7Ac8Cc7t0KMzPrbor4+J0OkrQ0MD8i5knaCvh9tWHMWrW0tERrq/9bwMxsYUiaHBEtXdlmM51T60prAX+R1At4FziyzvGYmVkX+FgmtYh4lPTvAWZmViJNcU7NzMysFk5qZmZWGk5qZmZWGk5qZmZWGk5qZmZWGk5qZmZWGk5qZmZWGk5qZmZWGh/Lf77uajNensGnL/50vcMwazrTDp1W7xCsZNxTMzOz0mjqpCbpKElfqXccZmbWGJp2+FHSkhFxTr3jMDOzxtHlPTVJh0i6W9IUSedKWlvSo5JWktRL0kRJu0gaLOkhSRdLul/SFZKWyW2MkHSrpMmSbpC0ei4fL+mnkm4Fvi1ptKTj8rwhksbmZSZKGpbLL5L0G0m3S3pc0v6FWL+f77s2VdJpHbVjZmaNr0uTmqRPAgcC2+T7k80HdgBOB84B/hN4ICJuzIsMBcZExMbA68A3JPUGfgvsHxEjgAuBUwurGRgRO0TE/1SsfgxwdF7mOODswrzVgW2BPYG25LUb8AXgMxGxCXBGDe0Ut3WUpFZJrfPnzK99J5mZWbfp6uHHzwIjgHskAfQFZkXEaEkjI2cUAAALIElEQVQHAEcBxZtxPhURk/LzPwLHAGOBjYCbchtLkO5M3eayypVK6gdsDVyelwFYulDlmohYADwgadVctjPwh4iYCxARr9TQzvsiYgwpAdJ3nb4fvzutmpk1oK5OagIujogffKgwDSuumSf7AXPy88pkELmNGRGxVTvreLNKWS9gdgd3r36nIsa2v5Xr76wdMzNrYF19Tu1mYH9JqwBIWkHS2qThx0uAk4DzCvXXktSWvL4I3AY8DKzcVi6pt6RPdbTSiHgdmJl7gyjZpJNYbwQOL5zHW2ER2zEzswbRpUktIh4ATgRulHQ/cBMwGNgcOD0iLgHelfTVvMiDwKG57grA7yPiXWB/4HRJU4EppCHBzhwMHJGXmQHs3UmsY4FrgVZJU0jnzxa6HTMzaxyKqM/pIEmDgesiYqO6BNCF+q7TN9YbvV69wzBrOv5FkY83SZMjoqUr22za/1NrJJ9a8VO0Htpa7zDMzD726pbUIuIJ0lWOZmZmXaKpfybLzMysyEnNzMxKw0nNzMxKw0nNzMxKw0nNzMxKw0nNzMxKw0nNzMxKw0nNzMxKw0nNzMxKwz+T1RWevQ9GD6h3FGaNbfRr9Y7APgbcUzMzs9Jo+qQm6TBJv6uhzhqF6fMlbdj90ZmZWU/6uAw/HgZMB54FiIiv1TUaMzPrFnXtqUm6RtJkSTMkjcplb0g6VdJUSXdKWjWXf17SXZLuk/TPtvJCW/0lzZTUO08vJ+mJfBfrFuASSVMk9ZU0XlJLrrerpHvz+m7OZTvkulPy+vr35H4xM7NFU+/hx8MjYgQp6RwjaUVgWeDOiNgEmAAcmeveBmwZEZsCfwa+X2woIuYA44E9ctFBwJURcTnQChwcEcMj4q22ZSStDJwH7JfXd0CedRzwzYgYDmwHvL9MYdlRkloltb44tz43WjUzsw+rd1I7RtJU4E5gELA+8C5wXZ4/GRicn68J3CBpGvA94FNV2jsf+Gp+/lXgD52sf0tgQkTMBIiIV3L5JOAXko4BBkbEvMoFI2JMRLRERMvKy6jTDTUzs+5Xt6QmaSSwM7BV7iXdB/QB3ouItq7PfD447/db4HcR8Wng67nuh0TEJGCwpB2AJSJiemdhAB/pZkXEacDXgL7AnZKGLeTmmZlZHdSzpzYAeDUi5uaksWUN9Z/Jzw/toN7/Apfy4V7aHKDaebE7gB0krQMgaYX8d0hETIuI00lDl05qZmZNoJ5JbSywpKT7gVNIQ5AdGQ1cLmki8FIH9S4BlicltjYXAee0XSjSVhgRLwKjgKvyMOhledaxkqbnsreAf9S8VWZmVjf6YKSvHCTtD+wdEV/uqXW2rLFEtI7q11OrM2tO/kURqyBpckS0dGWbpfo/NUm/BXYDdu/RFa+xKYxu7dFVmpnZR5UqqUXE0fWOwczM6qfel/SbmZl1GSc1MzMrDSc1MzMrDSc1MzMrDSc1MzMrDSc1MzMrDSc1MzMrDSc1MzMrjVL983W9THvmNQYff329wzDrdk+ctkfnlczqyD01MzMrDSc1MzMrjaZKapJGSrqu85rv1x8uaffC9F6Sju+e6MzMrN6aJqlJWpTzf8Mp/GJ/RFyb72ptZmYl1O1JTdJgSQ9JOj/fePMSSTtLmiTpUUlb5Mftku7Lf4fmZQ+TdLmkvwE3VrS7ea6/rqRlJV0o6Z5ctrekpYAfAwfmm4MemNv7XV7+Ikm/yet7PN+HDUm9JJ0taYak6yT9vW2emZk1tp7qqa0H/BrYGBgGfAnYFjgO+CHwELB9RGwKnAT8tLDsVsChEbFTW4GkrYFzSDcDfRw4AbglIjYHdgTOBHrnti6LiOERcRkftXqOY0+grQe3LzAY+DTwtbz+j5A0SlKrpNb5c33zQzOzRtBTl/TPjIhpAJJmADdHREiaRkogA4CLJa0PBCkhtbkpIl4pTH8SGAPsEhHP5rJdgL0kHZen+wBr1RDXNRGxAHhA0qq5bFvg8lz+vKRx1RaMiDE5DpZeff1y3T7czKxJ9VRSe6fwfEFhekGO4RRgXETsI2kwML5Q/82Ktp4jJa1NgbakJmC/iHi4WFHSZxYiLlX8NTOzJtMoF4oMAJ7Jzw/rpO5sYA/gp5JG5rIbgKMlCUDSprl8DtB/IWO5Ddgvn1tbFRjZSX0zM2sQjZLUzgB+JmkSsERnlSPiBeDzwFm5N3YKacjyfknT8zTAOGDDtgtFaozlSuBpYDpwLnAX4JNmZmZNQBE+HVRJUr+IeEPSisDdwDYR8Xx79VtaWqK1tbXnAjQzKwFJkyOipSvb9G8/VnedpIHAUsApHSU0MzNrHE5qVUTEyHrHYGZmC69RzqmZmZktNic1MzMrDSc1MzMrDV/92AUkzQEe7rRi41oJeKneQSyGZo6/mWMHx19vzR7/0IhY2P8l7pAvFOkaD3f1Zak9SVKr46+PZo4dHH+9lSH+rm7Tw49mZlYaTmpmZlYaTmpdY0y9A1hMjr9+mjl2cPz15vgr+EIRMzMrDffUzMysNJzUzMysNJzUFoOkXSU9LOlfko6vdzy1kHShpFn5Fj1tZStIuknSo/nv8vWMsT2SBkkaJ+lBSTMkfTuXN0v8fSTdLWlqjv9HuXwdSXfl+C+TtFS9Y22PpCUk3SfpujzdNLEDSHpC0rR8O6rWXNYsx89ASVdIeii/B7ZqotiH5n3e9nhd0rHdEb+T2iKStARwFrAbsCHwRUkb1jeqmlwE7FpRdjxwc0SsD9ycpxvRPOA/I+KTwJbAN/M+b5b43wF2iohNgOHArpK2BE4HfpnjfxU4oo4xdubbwIOF6WaKvc2OETG88P9dzXL8/BoYGxHDgE1Ir0NTxB4RD+d9PhwYAcwFrqY74o8IPxbhAWwF3FCY/gHwg3rHVWPsg4HphemHgdXz89VJ/0xe9zhr2I6/Ap9rxviBZYB7gc+QfhFiyVz+oeOqkR7AmvmDZyfgOkDNEnthG54AVqooa/jjB1gOmEm+uK+ZYq+yLbsAk7orfvfUFt0ngKcK00/nsma0akQ8B5D/rlLneDolaTCwKenO5E0Tfx6+mwLMAm4CHgNmR8S8XKWRj6NfAd8HFuTpFWme2NsEcKOkyZJG5bJmOH7WBV4E/pCHf8+XtCzNEXulg4BL8/Muj99JbdGpSpn/P6IHSOoHXAkcGxGv1zuehRER8yMNwawJbAF8slq1no2qc5L2BGZFxORicZWqDRd7hW0iYjPSaYNvStq+3gHVaElgM+D3EbEp8CYNOtTYkXzOdS/g8u5ah5PaonsaGFSYXhN4tk6xLK4XJK0OkP/OqnM87ZLUm5TQLomIq3Jx08TfJiJmA+NJ5wYHSmr7HdZGPY62AfaS9ATwZ9IQ5K9ojtjfFxHP5r+zSOd0tqA5jp+ngacj4q48fQUpyTVD7EW7AfdGxAt5usvjd1JbdPcA6+erv5YidamvrXNMi+pa4ND8/FDSuaqGI0nABcCDEfGLwqxmiX9lSQPz877AzqST/eOA/XO1how/In4QEWtGxGDSsX5LRBxME8TeRtKykvq3PSed25lOExw/EfE88JSkobnos8ADNEHsFb7IB0OP0B3x1/ukYTM/gN2BR0jnRU6odzw1xnwp8BzwHunb3xGkcyM3A4/mvyvUO852Yt+WNLx1PzAlP3Zvovg3Bu7L8U8HTsrl6wJ3A/8iDcssXe9YO9mOkcB1zRZ7jnVqfsxoe8820fEzHGjNx881wPLNEnuOfxngZWBAoazL4/fPZJmZWWl4+NHMzErDSc3MzErDSc3MzErDSc3MzErDSc3MzErDSc3MzErDSc3MzErj/wNHZGv+AH2kRQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Top 10 Common Words in \"NON\" Data Scientist Job Descriptions\n",
    "desc_freq=desc1_traindf.loc[df[\"is_datascientist\"] == 0].sum(axis=0)\n",
    "desc_freq.sort_values(ascending = False).iloc[0:10].sort_values().plot(kind='barh')\n",
    "plt.title('Top Common words in *Non* Data Scientist Job Descriptions')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Top 5 Common Words for Non Data Scientist Job Descriptions:\n",
    "- business, team, support, management, database"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZsAAAEICAYAAACJalkVAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvOIA7rQAAIABJREFUeJzt3XmYHFW9//H3hxAMSSAxQZB9EJAAggmMEQQCQkRULovAxQUloEZcQFRQrvhDEJWwXBVBkRARFEQFuRhBIBgMgUAgk4Us7ELY97BkgUCS7++Pc4YUTc8+NT2ZfF7P0093nzp1lurq+tapqu5SRGBmZlamNWrdADMz6/kcbMzMrHQONmZmVjoHGzMzK52DjZmZlc7BxszMSudgYz2SpDGSxrVjvv9I2rWMNnVXkt4v6eWSyn6XpEWSNiqj/GbqnSrpiK6sszNIulnS4e2ctybLurUcbJqRP7jGxwpJrxXef76E+j4l6TZJCyU9l1e8T3R2Pda0iNgyIu5o63yS+kgKSYvz+vGCpJskfboNZewn6aG21l1RxtckPZDb8Iykf0hau7l5IuKBiBjYkXoL9b9tIx8RSyOif0Q81cJ8Lfa9vTsQLcl1ryh8tx+XdIWkYZ1dV0siYu+I+Etr8rZ3WdeKg00z8gfXPyL6A48B/1VIu7wz68rB60/ARcDGwIbAT4EDO7OenkbSGpK603q8TV5ftgWuAMZJ+n5XVCzp48APgUNyG7YH/q8r6u4BHs7LbF3gI8AjwO2S9uiKyrvhetz5IsKPVjyA+cDIirS1gV8DTwNPAGcDvfO0/YCHgNOABcDDwGFNlL0m8AxwbDP198plPQY8C1wMrJOnDQGWAV8CngReBI4GdgXmAi8DPy+UdQxwM3A+8ArwIFAPjM7zPwt8ppB/ECkQPk/6En4PUKGsicCvcj3/qVxOhXK+BlxZeP848IfC++eAIfn1nsCM3L6pwIcK+aYCPwbuBF4HNgG2AqYAC4HrgQuBcTl/P+DP+XN4Oc/37iba+Aywe349BricFDQWArOBoU3M1wcIYJOK9COAJcC6+f1XgftyeQ8BR+f0wcBrwApgUX4MBnbL7X0FeAr4BbBmE234IfDnZtahfvlzejyXd0te94YAyyo+7z/kZfE48CNgjZY+b+B/geX5M1mU379tuZB2nhr7/zhwXFN9r9L+MY2faSvXkdOB6Xn634ABTSyX/YCHqqSPA24rvP8A6XvzEnAvcFBh2jv6VZh2WF53XiV91/ZpZj2eChxR8T29MM97DzCiDcu63d9b4Cukbd5Cmtl2tWkb2tECVpcH1YPNWcCtwHrABsA04OTCCrwMOANYCxhJ2uhsUaXsoXkl2bCZ+r+eV/DNSXtf1wIX5WlD8vznAu8CDgAW5y/YYGCz/AX5cGFFexP4HGljczbwKGlDtlae/yWgT87/V+BKoD9po/4I8PmKsr5ICojfBuY30YftgOfy6/flcuYXpj2bX6+fv1z/nds3Kn9hBhS+pA8D2wC9c54ZhWW9T17WjcHmW8BVpJ2DNYEPAf2aaGNlsFkCfCz37RfApCbmayrY9MvpH83vDwC2AJTXideA7QvrzEMV8w/P7e0FbEkKUMc00YbGdewU0o7GWhXTfwdMAN6by9sjP1cGm+uB84C+pBH2TODI1nzeFDaW1ZYLaUdoeH49GBjWVN+r9O+tYNPKdeTR3Lf+wD8oBKqKcpsKNp8kfYd7k75zTwOfz/3+EGnnZasW+rUH6bv0UdKRpM2A9zezHlcGm2Wk737vvMwXsHLHpaVl3a7vLfBuUgDaMr/fGNi2w9vQjhawujyoHmyeBPYuvD8QuK+wAr9O3mDntPHAiVXK3oe0V7dGM/VPIe8F5/cfJG1YxMpgM7gwfTFwYOH9deSNVF7R5hSmfSjPP6Bi/iGk4LUceF9h2reAGwplzS1MG5TLGthEP54jBZZRpL2q2UAdadTz15znK8Dkivlmkkdb+Uv2g8K091dZ1lezcsP0ddJe/Ada8TlXBptrC9N2Al5uYr6qwSZPe5l0aKvafDcAXy2sMy1tcE8Crmhm+gH5s341P84kbeR6kzYu21SZ561gQ9qZWUweoee0o4DrW/N50/IG8Nlc3joVbWhrsGnNOnJqxWe3uIlymwo2jTuBg4EjgZsqpl8KfL+Ffl0KnNFEvW9bjyuXX17Wj1RMn00eZTS3rOnA95aVweZACt+pjj569jHCEkkSaQ/x0ULyo6S9gEbPR8TrFdOrXSnyIilobNBMlRtVqWtt0koCsDwiXixMf430BSi+7194XzltaUS8UiX/e0kbq8cq6i7285nC6yX5uVhX0WRgL2AEKQBMIh0O2TO/h3f2tVqdjxdeb0T1Zd3od7nsqyQ9Ielnkno10b5KlX1rql9VSeoHDCDtkSLpAEl3SVqQrwDbmzQybmr+7SRdL+lZSa+SRi1N5o+I8RHxKdJG4zBSEP8CaYSyJmlPujmbkzZaz0t6ObfxXN6+brbl8650EHAI8Fi+AKa+lfNVaus68ijQV9KANtSxMWmD/SppuYxoXCZ5uRxCWq7QdL82JR2iasrjzUyDdHi+qKltSKV2f28j4iXSCO444BlJ4yVt1Yo6m+Vg006RdgeeIa2EjTYjjXYarSepT8X0aleKzCVt/A9ppsqnqtT1GnkjVqJnSKOuzSrqfrJ69hbdQgo2e5ACzy2kQNMYfOCdfa1WZxReP031ZZ0ypqt0TomIIbmew4DPtLP9bXUw6XOangPPlaRzCetHugLsZtKOBry9T40uIh0i3DIi1iUd41eVfG8TESsi4kbSMv4AaRktIx2+bM7jpHMA746IgfmxbkTs1FKdzfSh2K47ImJ/UvCaQDof1uJ8VbRmHdm0YtqSih2qlhwMTI2IN0nLZUJhmQyMdKHQ8dBsvx4nHf5sSkv93qTifXEb0ty8HfreRsR1EbEPKbA9BlzQmvma42DTMVcAP5I0WNL6wMnAZYXpvYH/J2ktSXuTjv3/rbKQiFgGnAD8RNIXJK2Tr07ZU9JvCnWdIGkzSesAPwH+lINeaSJiKemKpp9J6idpS9Jw/LLm52zSLaTDFm9ExPP5/SGkcy3zcp7xwDBJh0paU9IXSV+UG5oo8wHSydnGZf3RXAcAkkbmEcIapL3UZaQ91tLkdeJI4JfATyLiVdJItDfpUOIKSQeQAm+jZ4H1JRVHCesAr0TEIknbkw4fNVXnoZIOkzRQyUdIFxg0bjD/AJwraQNJvSTtXjnCi4hHSIdnziqsh1tL2r2VXX+WJgJaXn8+I2ld0iG9haz8HKr1vTmtWUdGKf2GqD9wKtDiJcV5uW0i6XTSxR0n50nX5PoOl9Q7r2e75PKb69c44KuSRuRluamk97eyjwCbSjom9/GI3McJeVqTy7oj31tJGyv9DKMvsJS089Hh74uDTcecQrpCZB4wi3Re5azC9PmkDdszpKvHjoqIqocxIuIy0sp9DGkv9BnSVUB/z1kuIJ2HuJ00LF8AfKdTe9O0r+bnR0l74uNIV2m1xxzSF3IyQES8QNr7u7UxcEbEs6RzDyeTDjF+E9g/Iqr+8DDPdzjpJOwC0lU3xS/VxqTluJA0ivwn6eRpGe6XtIgUAL8IfD0ifpbb+QJpp+IfpH4dlNvS6G7SRvTRfKhmEOnE7Zdzmb+m+Q3mS6TzU/8hBdWLgdMionEH57g8bWau/3Sqj5I+SzoMdx9pef6F5g/xFv0C+KKklySdVWX60aT16BXS8jkyp1frezVtWUf+SNpJe5K0l//dZtr9vryMF5GuDtuGdO7ullzfS8DHSedlniaNLn5C2nlosl8RcSvpO/2bPG0i7xytNGcyMIz0OZwMHFwYnbW0rNv7ve0F/A9pG/Qi6ZzusW1oc1Uqecd4tSVpP+D8iOjwsU4zgzzKfzUiTqp1W7qCpGOAQyNiZK3b0hk8sjGzbk/SYNJh6IZat8Xax8HGzLo1SYcA95MOBV1T4+ZYO/kwmpmZlc4jGzMzK92atW5Ad7HeeutFXV1drZthZrZKmT59+gsR8Z6W8jnYZHV1dTQ0+NyjmVlbSKr8J4eqfBjNzMxK52BjZmalc7AxM7PSOdiYmVnpfIFAdvfCJbz337Nq3Qwzsy71zEeHdkk9HtmYmVnpOjXY5H9NLVW++dRq8Ud8ZmY9Rbc8jCapV0RUvX9CRIwn/RW5mZmtIko7jCbpREnTJM2WdFoh/RpJ0yXNkzS6kL5I0o8l3QnsKmm+pNMkzZA0R9KQnG+UpPPz60sk/UrS7ZIelnRoTl9D0m9yHddK+mfjNDMz63qlBBtJ+wJbA8OBocDOkkbkyUdHxM5APXBc/utwgH7A3Ij4cETcltNeyLejvYB006lqNgR2B/YHxuS0TwN1wA7Al4Fdm2jnaEkNkhpWvFL1vlxmZtYJyhrZ7JsfM0n3Tx9CCj6QAszdpFvPblpIX847b5l8dX6eTgoe1VyT77d+DyvvJrg7cGVOfwb4d7UZI2JsRNRHRP0aAwa2pX9mZtYGZZ2zEXBGRFz4tkRpL2AksGtELJE0CeiTJ79e5TzN0vy8vJm2Li28VsWzmZl1A2WNbG4EjpbUH0DSxpLWBwYAL+VAMwTYpaT6bwMOyeduNgD2KqkeMzNrhVJGNhExQdK2wB2SABYBRwA3AMdImk26897UMuonHY7bB5gLPADcCbxSUl1mZtaCHnunTkn9I2JRvgDhLmC3fP6mqt7bbBeDf/unrmugmVk30NF/EJA0PSLqW8rXLX9n00mulTQQWAs4vblAA/DBdfrS0EV/22BmtrrpscEmIvaqdRvMzCzxf6OZmVnpHGzMzKx0DjZmZlY6BxszMyudg42ZmZXOwcbMzErnYGNmZqVzsDEzs9I52JiZWel67D8ItNXChXOYePOWtW6Gmdk77LP3f2rdhA7zyMbMzEpX82AjqV7Sr2rdDjMzK0/ND6NFRAPQUOt2mJlZeUob2UjqJ+k6SXdLmivpcEkfknR7TrtL0jqS9pJ0bWGeiyVNkzRT0oE5fZSkqyXdIOlBSWcV6tlP0oxc5sTmyjEzs9ooc2SzH/BURHwKQNIAYCZweERMk7Qu8FrFPCcDN0fE0fleNHdJ+leeNhQYBiwF7pd0HvA6cBEwIiIekTSouXIiYnGxMkmjgdEA669f80GemVmPVeY5mznASElnStoD2Ax4OiKmAUTEqxGxrGKefYGTJM0CJgF98nwAEyPilYh4HbgH2BzYBZgcEY/kMhe0opy3RMTYiKiPiPqBA2t++srMrMcqbXc+Ih6QtDPwSeAMYALQ0j2oBRwSEfe/LVH6MGlE02g5qe1qosyq5ZiZWW2Uec5mI2BJRFwGnEMahWwk6UN5+jqSKoPdjcCxkpTzDGuhmjuAPSVtkfM3HkZrazlmZlaiMk9U7ACcLWkF8CbwNdKI4zxJa5PO14ysmOd04JfA7Bwo5gP7N1VBRDyfz7tcLWkN4DngY20tx8zMyqWIlo5srR7q6+ujocFXYJuZtYWk6RFR31I+nxU3M7PSOdiYmVnpHGzMzKx0DjZmZlY6BxszMyudg42ZmZXOwcbMzErnYGNmZqVzsDEzs9I52JiZWel8E5fsqaee4tRTT611M8ysi/j73rU8sjEzs9I52JiZWek6HGwk1Uma28EyNpJ0VUfbYmZm3VO3OGcTEU8Bh9a6HWZmVo7OOoy2pqRLJc2WdJWkvpLmS1oPQFK9pEn59Z6SZuXHzHzHzrdGR5JGSbpa0g2SHpR0VmMlkvaVdIekGZKulNQ/p4+RdE+u/5ycdpikuZLuljS5k/ppZmbt0Fkjm22AL0XEFEkXA19vJu8JwDdy3v7A61XyDAWGAUuB+yWdR7qz5w+BkRGxWNL3ge9IOh84GBgSESFpYC7jFODjEfFkIe1t8l0+RwMMGDCgrX02M7NW6qyRzeMRMSW/vgzYvZm8U4CfSzoOGBgRy6rkmRgRr0TE68A9wObALsB2wBRJs4Ajc/qrpIA1TtKngSWFei6R9BWgV7WGRMTYiKiPiPq+ffu2pb9mZtYGnRVsKu8tHcCyQvl93poQMQb4MrA2MFXSkCrlLS28Xk4agQm4KSKG5sd2EfGlHKyGA38DDgJuyPUcQxoJbQrMkjS4g300M7N26qxgs5mkXfPrzwK3AfOBnXPaIY0ZJW0ZEXMi4kygAagWbKqZCuwmaatcTl9J78+H4gZExD+B40mH4BrruTMiTgFeIAUdMzOrgc46Z3MvcKSkC4EHgQuAu4DfSfoBcGch7/GSPkoasdwDXA9s2FIFEfG8pFHAFZLelZN/CCwE/i6pD2n08+087WxJW+e0icDdHeuimZm1lyIqj4Ctnurr66OhoaHWzTAzW6VImh4R9S3l8z8ImJlZ6RxszMysdA42ZmZWOgcbMzMrnYONmZmVzsHGzMxK52BjZmalc7AxM7PSOdiYmVnpHGzMzKx03eJOnd3BG08u4omTbq11M8ysBJuM2aPWTVjteWRjZmalW6WCTfFW0x3JY2ZmXWuVCjZmZrZqKj3YSKqTdJ+kcZLmSrpc0khJUyQ9KGm4pEGSrpE0W9JUSTvmeQdLmiBpZr5XjgrlHiHpLkmzJF0oqVdFvf0kXSfp7lzv4WX31czMquuqkc1WwLnAjqQ7c34O2B04AfgBcBowMyJ2zO//kOf7EXBbRAwDxgObAUjaFjgc2C0ihpJuxPb5ijr3A56KiA9GxAfIt4sukjRaUoOkhgVLXu7M/pqZWUFXBZtH8q2gVwDzgImR7to2B6gjBZ4/AkTEzcBgSQOAEcBlOf064KVc3j6kW05PkzQrv39fRZ1zgJGSzpS0R0S8UtmoiBgbEfURUT+o78DO7bGZmb2lqy59Xlp4vaLwfkVuw7Iq80TFc5GASyPif5qqMCIekLQz8EngDEkTIuLHbW65mZl1WHe5QGAy+TCYpL2AFyLi1Yr0TwDvzvknAodKWj9PGyRp82KBkjYClkTEZcA5wE5d0A8zM6uiu/yo81Tg95JmA0uAI3P6acAVkmYAtwCPAUTEPZJ+CEyQtAbwJvAN4NFCmTsAZ0takad/rSs6YmZm76R06sR23HBI/PPIi2rdDDMrgf9BoDySpkdEfUv5usvIpubW2ri/V0gzs5J0l3M2ZmbWgznYmJlZ6RxszMysdA42ZmZWOgcbMzMrnYONmZmVzsHGzMxK52BjZmalc7AxM7PSOdiYmVnp/Hc12bMPP8T/Hr5/rZthZh303b9cW+smWBUe2ZiZWem6bbCRNF/SelXSD5B0Ui3aZGZm7bPKHUaLiPHA+Fq3w8zMWq9TRjaS6iTdJ2mcpLmSLpc0UtIUSQ9KGp4ft0uamZ+3yfP2knSOpDmSZks6tlD0sZJm5GlDcv5Rks7Pry+R9Ktc3sOSDi206URJ03KZp3VGP83MrH068zDaVsC5wI7AEOBzwO7ACcAPgPuAERExDDgF+FmebzSwBTAsInYELi+U+UJE7ARckMupZsNcz/7AGABJ+wJbA8OBocDOkkZUzihptKQGSQ2Ll77R3n6bmVkLOvMw2iMRMQdA0jxgYkSEpDlAHTAAuFTS1kAAvfN8I4HfRsQygIhYUCjz6vw8Hfh0E/VeExErgHskbZDT9s2Pmfl9f1LwmVycMSLGAmMBNh000LcsNTMrSWcGm6WF1ysK71fkek4H/h0RB0uqAybl6SIFn+bKXN5MW4v1qvB8RkRc2Mq2m5lZibryarQBwJP59ahC+gTgGElrAkga1Al13QgcLal/LnNjSet3QrlmZtYOXRlszgLOkDQF6FVIHwc8BsyWdDfpXE+HRMQE4E/AHfkw3lXAOh0t18zM2kcRPlUB6ZzN8R/bvdbNMLMO8j8IdC1J0yOivqV8q9zvbMqywfu28kpqZlaSbvsPAmZm1nM42JiZWekcbMzMrHQONmZmVjoHGzMzK52DjZmZlc7BxszMSudgY2ZmpXOwMTOz0jnYmJlZ6fx3Ndlzjy7k18fcXOtmmHV73/jt3rVugq2CPLIxM7PSdYtgI2mUpPNbkWejwvtxkrYrv3VmZtZRq9JhtFHAXOApgIj4ck1bY2ZmrdbhkY2kayRNlzRP0uictkjSTyXdLWmqpA1y+n9JulPSTEn/akwvlLWOpEck9c7v15U0X9JhQD1wuaRZktaWNElSfc63n6QZub6JOW3PnHdWrs83TzMzq5HOOIx2dETsTAoGx0kaDPQDpkbEB4HJwFdy3tuAXSJiGPBn4HvFgiJiITAJ+FRO+gzwt4i4EmgAPh8RQyPitcZ5JL0HuAg4JNd3WJ50AvCNiBgK7AG8NU9h3tGSGiQ1LHr95Y4uBzMza0JnBJvj8u2cpwKbAlsDbwCNdyKbDtTl15sAN+ZbNZ8IbF+lvHHAUfn1UcDvW6h/F2ByRDwCEBELcvoU4OeSjgMGRsSyyhkjYmxE1EdEff8+A1vsqJmZtU+Hgo2kvYCRwK55VDET6AO8GSvvN72cleeGzgPOj4gdgK/mvG8TEVOAOkl7Ar0iYm5LzQDecW/riBgDfBlYG5gqaUgbu2dmZp2koyObAcBLEbEkb8x3aUX+J/PrI5vJ9wfgCt4+qlkIVDvvcgewp6QtACQNys9bRsSciDiTdAjOwcbMrEY6GmxuANaUNBs4nXQorTmnAldKuhV4oZl8lwPvJgWcRpcAv228QKAxMSKeB0YDV+fDeX/Jk46XNDenvQZc3+pemZlZp9LKo13dh6RDgQMj4gtdVWd9fX00NDR0VXVmZj2CpOkRUd9Svm73OxtJ5wGfAD5Z67aYmVnn6HbBJiKOrXUbzMysc3WLv6sxM7OezcHGzMxK52BjZmalc7AxM7PSOdiYmVnpHGzMzKx0DjZmZlY6BxszMytdt/tRZ628Pnce9w7ZttbNMOuwbe+7t9ZNMHsHj2zMzKx0DjZmZla6Lgs2kiZJavafQSUdJGm7wvsfSxpZfuvMzKxM3W1kcxDwVrCJiFMi4l81bI+ZmXWCjt4Wup+k6yTdnW9UdrikfSTNlDRH0sWS3lVlvkWF14dKukTSR4ADgLPzDdK2zOmH5nxVy5U0X9JpkmbkaUNy+p65nFl5vmp3+TQzsy7Q0ZHNfsBTEfHBiPgA6c6dlwCHR8QOpKvdvtaagiLidmA8cGJEDI2I/zROk9SnhXJfiIidgAuAE3LaCcA3ImIosAfpbp1vI2m0pAZJDQuWL2tDt83MrC06GmzmACMlnSlpD6AOeCQiHsjTLwVGdLAOgG1aKPfq/Dw9twFgCvBzSccBAyPiHdEkIsZGRH1E1A/q5avAzczK0qFgkzf+O5OCzhnAga2dtfC6Tyvyq4XpS/PzcvJvhyJiDPBlYG1gauPhNTMz63odPWezEbAkIi4DzgE+AtRJ2ipn+QJwS5VZn5W0raQ1gIML6QuBaudW7mtlucW2bRkRcyLiTKABcLAxM6uRjh472oF0Qn8F8CbpPMoA4EpJawLTgN9Wme8k4FrgcWAu0D+n/xm4KB/6OrQxc0S8LumoVpRbdLykj5JGO/cA17evi2Zm1lGKiJZzrQbq6+ujoaGh1s0wM1ulSJoeEc3+hhK63+9szMysB3KwMTOz0jnYmJlZ6RxszMysdA42ZmZWOgcbMzMrnYONmZmVzsHGzMxK52BjZmalc7AxM7PS+X/1s3kvzmOHS3eodTOsh5tz5JxaN8GsJjyyMTOz0nWbYCNpoKSv17odZmbW+bpNsAEGAg42ZmY9UHcKNmOALSXNknS2pBMlTZM0W9JpjZkkXSNpuqR5kkYX0hfl21NPl/QvScMlTZL0sKQDatIjMzMDulewOQn4T0QMBW4CtgaGA0OBnSWNyPmOjoidgXrgOEmDc3o/YFKethD4CfAx0p1Af1ytQkmjJTVIali+cHlZ/TIzW+1116vR9s2Pmfl9f1LwmUwKMI23kt40p78IvAHckNPnAEsj4k1Jc4C6apVExFhgLMDaW6ztu8iZmZWkuwYbAWdExIVvS5T2AkYCu0bEEkmTgD558pux8rajK4ClABGxIt9K2szMaqQ7HUZbCKyTX98IHC2pP4CkjSWtDwwAXsqBZgiwS22aamZmbdFt9vgj4kVJUyTNBa4H/gTcIQlgEXAE6TDZMZJmA/cDU2vVXjMza71uE2wAIuJzFUnnVsn2iSbm7V94fWpT08zMrOt1q2BTS9sP3p6GIxtq3Qwzsx6pO52zMTOzHsrBxszMSudgY2ZmpXOwMTOz0jnYmJlZ6RxszMysdA42ZmZWOgcbMzMrnYONmZmVzsHGzMxK57+rafTUTDh1QK1bYauaU1+pdQvMVgke2ZiZWem6JNhI+kFX1GNmZt1TV41sujTY+M6cZmbdS6uCjaQvSpot6W5Jf5R0iaRDC9MX5ecNJU2WNEvSXEl7SBoDrJ3TLs/5vpOnz5V0fE6rk3SfpHE5/XJJI/MN1R6UNDzn6yfpYknTJM2UdGBOHyXpSkn/ACZUa0vnLjozM2utFkcAkrYHTgZ2i4gXJA0Cft5E9s8BN0bETyX1AvpGxK2SvhkRQ3N5OwNHAR8GBNwp6RbgJWAr4DBgNDAtl7c7cABpdHRQbsvNEXG0pIHAXZL+levfFdgxIhZI+m5lW6r0bXSui80GqKVFYWZm7dSaw017A1dFxAsAeUPeVN5pwMWSegPXRMSsKnl2B/4vIhYDSLoa2AMYDzwSEXNy+jxgYkSEpDlAXZ5/X+AASSfk932AzfLrmyJiQWvbEhFjgbEA9Rv1ipYXhZmZtUdrDqMJqNwQL2ucVynyrAUQEZOBEcCTwB8lfbGJ8pqytPB6ReH9ClYGRgGHRMTQ/NgsIu7N0xY3ztzKtpiZWRdoTbCZCPy3pMEA+TDafGDnPP1AoHeetjnwXERcBPwO2CnneTOPMAAmAwdJ6iupH3AwcGsb2nwjcGwOckgaVi1TM20xM7Mu1uJhtIiYJ+mnwC2SlgMzge8Df5d0FykYNY4o9gJOlPQmsAhoHE2MBWZLmhERn5d0CXBXnjYuImZKqmtlm08HfpnLEynw7V8lX1NtMTOzLqYIn6qAdM6mYXT/WjfDVjX+BwFbzUmaHhH1LeXz71EabTQMTm2odSvMzHok/12NmZmVzsHGzMxK52BjZmalc7AxM7PSOdiYmVnpHGzMzKx5XxquAAAFyUlEQVR0DjZmZlY6BxszMyudg42ZmZXO/yCQzXnyFepOuq7WzbBuaP6YT9W6CWarPI9szMysdA42ZmZWuh4ZbCRNktTiv5CamVnX6NRgo6RLApgkn28yM1tFdDgwSKqTdK+k3wAzgC9IukPSDElXSuqf842RdI+k2ZLOyWnvkfQ3SdPyY7ecPlzS7ZJm5udtcvqoXOY/gAk57XuS5ki6W9KYQtMOk3SXpAck7dHRfpqZWft11uhgG+Ao4BTgamBkRCyW9H3gO5LOJ93+eUhEhKSBeb5zgV9ExG2SNiPd8nlb4D5gREQskzQS+BlwSJ5nV2DHiFgg6RPAQcCHI2JJvmX1W32LiOGSPgn8CBhZ2WhJo4HRAL3WfU8nLQozM6vUWcHm0YiYKml/YDtgSrpjM2sBdwCvAq8D4yRdB1yb5xsJbJfzAqwraR1gAHCppK2BAHoX6ropIhYU5v99RCwBKKRDCnoA04G6ao2OiLGkW1bzrg239i1LzcxK0lnBZnF+FikYfLYyg6ThwD7AZ4BvAnuTDuPtGhGvVeQ9D/h3RBwsqQ6YVKWuxvqaChJL8/Ny/HsiM7Oa6uyT+VOB3SRtBSCpr6T35/M2AyLin8DxwNCcfwIp8JDzN6YPAJ7Mr0c1U98E4GhJffP8g5rJa2ZmNdKpwSYinicFhyskzSYFnyHAOsC1Oe0W4Nt5luOA+nzRwD3AMTn9LOAMSVOAXs3UdwMwHmiQNAs4oTP7Y2ZmnUMRPlUBUF9fHw0NDbVuhpnZKkXS9Iho8XeNPfJHnWZm1r042JiZWekcbMzMrHQONmZmVjoHGzMzK52vRsskLQTur3U7ush6wAu1bkQXWp366772TN25r5tHRIv/9+Vf1q90f2su3+sJJDWsLn2F1au/7mvP1BP66sNoZmZWOgcbMzMrnYPNSmNr3YAutDr1FVav/rqvPdMq31dfIGBmZqXzyMbMzErnYGNmZqVzsAEk7SfpfkkPSTqp1u0pi6RNJf1b0r2S5kn6Vq3bVDZJvSTNlHRty7lXXZIGSrpK0n3589211m0qi6Rv5/V3rqQrJPWpdZs6k6SLJT0naW4hbZCkmyQ9mJ/fXcs2tsdqH2wk9QJ+DXyCdEvrz0rarratKs0y4LsRsS2wC/CNHtzXRt8C7q11I7rAucANETEE+CA9tM+SNibfBysiPkC639VnatuqTncJsF9F2knAxIjYGpiY369SVvtgAwwHHoqIhyPiDeDPwIE1blMpIuLpiJiRXy8kbZA2rm2ryiNpE+BTwLhat6VMktYFRgC/A4iINyLi5dq2qlRrAmtLWhPoCzxV4/Z0qoiYDCyoSD4QuDS/vhQ4qEsb1QkcbNLG9vHC+yfowRvgRpLqgGHAnbVtSal+CXwPWFHrhpTsfcDzwO/zIcNxkvrVulFliIgngXOAx4CngVciYkJtW9UlNoiIpyHtNALr17g9beZgA6qS1qOvB5fUH/gbcHxEvFrr9pRB0v7AcxExvdZt6QJrAjsBF0TEMGAxq+BhltbI5yoOBLYANgL6STqitq2y1nCwSSOZTQvvN6GHDcuLJPUmBZrLI+LqWrenRLsBB0iaTzo0ureky2rbpNI8ATwREY2j1KtIwacnGgk8EhHPR8SbwNXAR2rcpq7wrKQNAfLzczVuT5s52MA0YGtJW0hai3SycXyN21QKSSId1783In5e6/aUKSL+JyI2iYg60md6c0T0yD3giHgGeFzSNjlpH+CeGjapTI8Bu0jqm9fnfeihF0NUGA8cmV8fCfy9hm1pl9X+X58jYpmkbwI3kq5suTgi5tW4WWXZDfgCMEfSrJz2g4j4Zw3bZJ3jWODyvMP0MHBUjdtTioi4U9JVwAzS1ZUz6QF/5VIk6QpgL2A9SU8APwLGAH+V9CVSwD2sdi1sH/9djZmZlc6H0czMrHQONmZmVjoHGzMzK52DjZmZlc7BxszMSudgY2ZmpXOwMTOz0v1/TgGo1GF2HysAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Top 10 Common Words in Data Scientist Job Descriptions\n",
    "desc_freq = desc1_traindf.loc[df[\"is_datascientist\"] == 1].sum(axis=0)\n",
    "desc_freq.sort_values(ascending = False).iloc[0:10].sort_values().plot(kind='barh')\n",
    "plt.title('Top Common words in Data Scientist Job Descriptions')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Top 5 Common Words for Data Scientist Job Descriptions:\n",
    "- learning, science, business, models, machine, analytics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Modelling Time:\n",
    "# To use Logistic Regression, Random Forest, KNearest models to see their performance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'C': 3, 'penalty': 'l1'}\n",
      "                precision    recall  f1-score   support\n",
      "\n",
      "        Others       0.95      0.97      0.96       392\n",
      "Data Scientist       0.38      0.25      0.30        24\n",
      "\n",
      "     micro avg       0.93      0.93      0.93       416\n",
      "     macro avg       0.67      0.61      0.63       416\n",
      "  weighted avg       0.92      0.93      0.93       416\n",
      "\n",
      "ROC Score: 0.8601721938775511\n"
     ]
    }
   ],
   "source": [
    "# Using Logistic Regression\n",
    "lr = LogisticRegression(solver='liblinear', random_state=91)\n",
    "C_vals = [.0001,.001,.01,.05, .1,.12,.13,.135,.15,.16,.17,1,3,5,10,20,50,100]\n",
    "penalties = ['l1','l2']\n",
    "\n",
    "# Using GridSearch to find best penalty and C parameters\n",
    "gs = GridSearchCV(lr, {'penalty': penalties, 'C': C_vals}, verbose=False, cv=5)\n",
    "gs.fit(desc1_train, y_train)\n",
    "print(gs.best_params_)\n",
    "y_pred_lr6 = gs.best_estimator_.predict(desc1_test)\n",
    "y_predp_lr6 = gs.best_estimator_.predict_proba(desc1_test)[:, 1]\n",
    "print(classification_report(y_test, y_pred_lr6, target_names=[\"Others\",\"Data Scientist\"]))\n",
    "roc_lr6 = roc_auc_score(y_test, y_predp_lr6)\n",
    "print(\"ROC Score:\", roc_lr6)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                precision    recall  f1-score   support\n",
      "\n",
      "        Others       0.95      0.99      0.97       392\n",
      "Data Scientist       0.60      0.12      0.21        24\n",
      "\n",
      "     micro avg       0.94      0.94      0.94       416\n",
      "     macro avg       0.77      0.56      0.59       416\n",
      "  weighted avg       0.93      0.94      0.93       416\n",
      "\n",
      "ROC Score: 0.8056972789115646\n"
     ]
    }
   ],
   "source": [
    "# Using Random Forest\n",
    "rf = RandomForestClassifier(random_state=91)\n",
    "rf.fit(desc1_train, y_train)\n",
    "y_pred_rf6 = rf.predict(desc1_test)\n",
    "y_predp_rf6 = rf.predict_proba(desc1_test)[:,1]\n",
    "print(classification_report(y_test, y_pred_rf6, target_names=[\"Others\",\"Data Scientist\"]))\n",
    "roc_rf6 = roc_auc_score(y_test, y_predp_rf6)\n",
    "print(\"ROC Score:\", roc_rf6)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                precision    recall  f1-score   support\n",
      "\n",
      "        Others       0.96      0.99      0.97       392\n",
      "Data Scientist       0.55      0.25      0.34        24\n",
      "\n",
      "     micro avg       0.94      0.94      0.94       416\n",
      "     macro avg       0.75      0.62      0.66       416\n",
      "  weighted avg       0.93      0.94      0.93       416\n",
      "\n",
      "ROC Score: 0.8256271258503401\n"
     ]
    }
   ],
   "source": [
    "# Using KNearest Neighbours\n",
    "knn = KNeighborsClassifier(n_neighbors=4)\n",
    "knn.fit(desc1_train, y_train)\n",
    "y_pred_kn6 = knn.predict(desc1_test)\n",
    "y_predp_kn6 = knn.predict_proba(desc1_test)[:,1]\n",
    "print(classification_report(y_test, y_pred_kn6, target_names=[\"Others\",\"Data Scientist\"]))\n",
    "roc_kn6 = roc_auc_score(y_test, y_predp_kn6)\n",
    "print(\"ROC Score:\", roc_kn6)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x1728 with 3 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(3, 1, figsize=[8,24])\n",
    "\n",
    "# Logistic Regression ROC Plot\n",
    "plt.subplot(311)\n",
    "fpr, tpr, _ = roc_curve(y_test, y_predp_lr6)\n",
    "roc_auc = auc(fpr, tpr)\n",
    "plt.plot(fpr, tpr, label='ROC curve (area = %0.2f)' % roc_auc, linewidth=4)\n",
    "plt.plot([0, 1], [0, 1], 'k--', linewidth=4)\n",
    "plt.xlim([-0.05, 1.05])\n",
    "plt.ylim([-0.05, 1.05])\n",
    "plt.xlabel('False Positive Rate\\n(1 - Specificity)', fontsize=18)\n",
    "plt.ylabel('True Positive Rate\\n(Sensitivity)', fontsize=18)\n",
    "plt.title('ROC for Logistic Regression', fontsize=18)\n",
    "plt.legend(loc=\"best\")\n",
    "\n",
    "# Random Forest ROC Plot\n",
    "plt.subplot(312)\n",
    "fpr, tpr, _ = roc_curve(y_test, y_predp_rf6)\n",
    "roc_auc = auc(fpr, tpr)\n",
    "plt.plot(fpr, tpr, label='ROC curve (area = %0.2f)' % roc_auc, linewidth=4)\n",
    "plt.plot([0, 1], [0, 1], 'k--', linewidth=4)\n",
    "plt.xlim([-0.05, 1.05])\n",
    "plt.ylim([-0.05, 1.05])\n",
    "plt.xlabel('False Positive Rate\\n(1 - Specificity)', fontsize=18)\n",
    "plt.ylabel('True Positive Rate\\n(Sensitivity)', fontsize=18)\n",
    "plt.title('ROC for Random Forest', fontsize=18)\n",
    "plt.legend(loc=\"best\")\n",
    "\n",
    "# KNearest ROC Plot\n",
    "plt.subplot(313)\n",
    "fpr, tpr, _ = roc_curve(y_test, y_predp_kn6)\n",
    "roc_auc = auc(fpr, tpr)\n",
    "plt.plot(fpr, tpr, label='ROC curve (area = %0.2f)' % roc_auc, linewidth=4)\n",
    "plt.plot([0, 1], [0, 1], 'k--', linewidth=4)\n",
    "plt.xlim([-0.05, 1.05])\n",
    "plt.ylim([-0.05, 1.05])\n",
    "plt.xlabel('False Positive Rate\\n(1 - Specificity)', fontsize=18)\n",
    "plt.ylabel('True Positive Rate\\n(Sensitivity)', fontsize=18)\n",
    "plt.title('ROC for Random Forest', fontsize=18)\n",
    "plt.legend(loc=\"best\")\n",
    "\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th>Summary:</th>\n",
       "      <th>LogisticReg</th>\n",
       "      <th>RandomForest</th>\n",
       "      <th>KNearest</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>ROC AUC Score</th>\n",
       "      <td>0.860172</td>\n",
       "      <td>0.805697</td>\n",
       "      <td>0.825627</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Baseline</th>\n",
       "      <td>0.934249</td>\n",
       "      <td>0.934249</td>\n",
       "      <td>0.934249</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "Summary:       LogisticReg  RandomForest  KNearest\n",
       "ROC AUC Score     0.860172      0.805697  0.825627\n",
       "Baseline          0.934249      0.934249  0.934249"
      ]
     },
     "execution_count": 79,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.DataFrame([[roc_lr6, roc_rf6, roc_kn6],\n",
    "              [baseline6, baseline6, baseline6]],\n",
    "            columns=[\"LogisticReg\", \"RandomForest\", \"KNearest\"],\n",
    "            index=[\"ROC AUC Score\", \"Baseline\"]).rename_axis('Summary:', axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Conclusion:\n",
    "- Logistic Regression Model was the better model to classify if the job title included Data Scientist or not, based on ROC Score.\n",
    "- As the number of job titles containing Data Scientist was much lesser than that of non Data Scientists, it resulted in a unbalanced class distribution.\n",
    "- Due to above, we should consider ROC score instead of Accuracy score to determine the model's performance."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### What components of a job posting distinguish data scientists from other data jobs? "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 158,
   "metadata": {},
   "outputs": [],
   "source": [
    "X7 = df[df[\"is_datascientist\"] == 1]['Job Description']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 159,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[nltk_data] Downloading package stopwords to\n",
      "[nltk_data]     C:\\Users\\vince\\AppData\\Roaming\\nltk_data...\n",
      "[nltk_data]   Package stopwords is already up-to-date!\n"
     ]
    }
   ],
   "source": [
    "# **********   NLP - Topic Modelling (First Part Only) **********  \n",
    "\n",
    "# Using job description we can identify main component of a job posting distinguish data scientists from other data jobs\n",
    "\n",
    "from gensim import corpora, models, matutils\n",
    "from sklearn.feature_extraction.text import CountVectorizer, TfidfVectorizer\n",
    "from sklearn.feature_extraction.text import ENGLISH_STOP_WORDS\n",
    "from collections import defaultdict\n",
    "\n",
    "#### Using STOP WORDS #### \n",
    "\n",
    "from nltk.corpus import stopwords\n",
    "import nltk\n",
    "\n",
    "# Use stop word from NLTK\n",
    "\n",
    "nltk.download('stopwords')\n",
    "nltk_stops = stopwords.words()\n",
    "\n",
    "# Use stop word from SKLearn\n",
    "custom_stop_words = list(ENGLISH_STOP_WORDS)\n",
    "custom_stop_words.append('job')\n",
    "custom_stop_words.append(\"role\")\n",
    "custom_stop_words.append(\"work\")\n",
    "\n",
    "# Fit and Transform using sklearn stop word.\n",
    "vectorizer = CountVectorizer(stop_words=custom_stop_words)\n",
    "X = vectorizer.fit_transform(X7)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 160,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "data         534\n",
       "business     188\n",
       "learning     169\n",
       "analytics    129\n",
       "machine      118\n",
       "team         116\n",
       "solutions    110\n",
       "models       102\n",
       "science       95\n",
       "develop       65\n",
       "dtype: int64"
      ]
     },
     "execution_count": 160,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Use Vocabulary - parameter of CountVectorizer\n",
    "# `.vocabulary_` attribute of the vectorizer contains a dictionary of terms. \n",
    "\n",
    "vectorizer.vocabulary_\n",
    "vectorizer.get_feature_names()\n",
    "\n",
    "# Get counts of tokens.\n",
    "docs = pd.DataFrame(X.todense(), \n",
    "                    columns=vectorizer.get_feature_names())\n",
    "docs.sum().sort_values(ascending=False).head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learning, science, business, models, machine, analytics"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### From job description, we can conclude the word  \"data\", \"business\", \"learning\", \"analytics\", \"machine\"  is a major factor for data scientist related job. These words are almost in line with the earlier result  under section \"Top 5 Common Words for Data Scientist Job Descriptions\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "-----"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### What features are important for distinguishing junior vs. senior positions?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 170,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Using job requirement, we can identify features that are important for distinguishing senior vs junior position\n",
    "X8 = df[df[\"Senior Management\"] == 1]['Job Requirements']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 173,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "experience     167\n",
       "data           141\n",
       "skills          84\n",
       "business        79\n",
       "development     72\n",
       "management      60\n",
       "years           55\n",
       "strong          54\n",
       "knowledge       53\n",
       "ability         53\n",
       "dtype: int64"
      ]
     },
     "execution_count": 173,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Fit and Transform using sklearn stop word.\n",
    "vectorizer = CountVectorizer(stop_words=custom_stop_words)\n",
    "X = vectorizer.fit_transform(X8)\n",
    "docs = pd.DataFrame(X.todense(), \n",
    "                    columns=vectorizer.get_feature_names())\n",
    "docs.sum().sort_values(ascending=False).head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### From job requirements' counts of tokens, it is logical to conclude that more \"Experience\",  \"Data\", \"Skills\", \"Business\", \"Management\" would be the deciding factor between a senior roles and non senior roles. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "-----"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Do the requirements for titles vary significantly with industry (e.g. healthcare vs. government)?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 195,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "experience    16\n",
       "ensure         7\n",
       "clinical       6\n",
       "medical        6\n",
       "degree         6\n",
       "maintain       5\n",
       "local          5\n",
       "assigned       5\n",
       "relevant       5\n",
       "design         5\n",
       "dtype: int64"
      ]
     },
     "execution_count": 195,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Using job requirement, we can identify features that are important for distinguishing senior vs junior position\n",
    "X9 = df[df[\"Healthcare / Pharmaceutical\"] == 1]['Job Requirements']\n",
    "# Fit and Transform using sklearn stop word.\n",
    "vectorizer = CountVectorizer(stop_words=custom_stop_words)\n",
    "X = vectorizer.fit_transform(X9)\n",
    "docs = pd.DataFrame(X.todense(), \n",
    "                    columns=vectorizer.get_feature_names())\n",
    "docs.sum().sort_values(ascending=False).head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 196,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "microsoft     2\n",
       "nitec         2\n",
       "office        2\n",
       "proficient    2\n",
       "www           1\n",
       "gce           1\n",
       "higher        1\n",
       "good          1\n",
       "english       1\n",
       "inform        1\n",
       "dtype: int64"
      ]
     },
     "execution_count": 196,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Using job requirement, we can identify features that are important for distinguishing senior vs junior position\n",
    "X10 = df[df[\"Public / Civil Service\"] == 1]['Job Requirements']\n",
    "# Fit and Transform using sklearn stop word.\n",
    "vectorizer = CountVectorizer(stop_words=custom_stop_words)\n",
    "X = vectorizer.fit_transform(X10)\n",
    "docs = pd.DataFrame(X.todense(), \n",
    "                    columns=vectorizer.get_feature_names())\n",
    "docs.sum().sort_values(ascending=False).head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### From first glance, \"Healthcare\" jobs has higher qualification requirement in the area of clinical and medical sector, where as governement jobs requirement is GCE level only. \n",
    "\n",
    "##### But this is inconclusive as there is insufficient data on \"Public / Civil Service\". Many job requirement in this sector is blank (null). This is because most government jobs will be forwarded to their respective gov/corporate websites. Most investigation is needed. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "-----"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "vocab = [\"r\", \"c++\", \"java\", \"javascript\", \"python\", \"alteryx\", \"analytics\", \"visualization\", \"mysql\", \"sqlite\",\n",
    "         \"sql\", \"tableau\", \".net\", \"c#\", \"etl\", \"powerbi\", \"machine\", \"learning\", \"informatica\", \"oracle\", \"rds\",\n",
    "         \"stata\", \"edm\", \"xml\", \"sas\", \"sybase\", \"sap\", \"datawarehouse\", \"data warehouse\", \"aws\", \"acess\", \"redis\",\n",
    "         \"excel\", \"hadoop\", \"hive\", \"spark\", \"scala\", \"nosql\", \"nlp\", \"azure\", \"netezza\", \"gcp\", \"Informatix\", \"DB2\",\n",
    "         \"Robomongo\", \"mapreduce\", \"rdbms\", \"matlab\", \"mssql\", \"mongodb\", \"predictive\", \"model\", \"teradata\", \"PostgreSQL\",\n",
    "         \"powerpoint\", \"sharepoint\", \"cognos\", \"business objects\", \"database\", \"data store\", \"couchdb\"]"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}