Machine Learning with Python

Welcome to my repository showcasing the projects I completed as part of the freeCodeCamp "Machine Learning with Python" course. These projects cover a range of machine learning concepts and applications. Feel free to explore each project's folder for more details.

Projects

I. Rock, Paper, Scissors Challenge

• Description: Introduction to Reinforcement Learning, by implementing the Q-Learning algorithm.

• Files:

  • rps_game.py: Provides boilerplate code containing four different opponent strategies to compete with our agent.
  • rps_agent.py: Our agent implementation via a QLearning class.
  • rps_agent_ens.py: Modified agent to be used in an ensemble model.
  • test_script.ipynb: Simulating our bot competting against the others, displaying its win rate. Additionally, an alternative method that incorporates several instances of the agent (ensemble) is presented.

II. Cat and Dog Image Classifier

• Description: Implementation of a CNN that correctly classifies images of cats and dogs.

• Files:

  • cat_dog_classification.ipynb: The training dataset undergoes augmentation through random transformations applied to the existing samples. A comparison is conducted between the deployed CNN and a pre-trained MobileNetV2 model to which a classifier layer is added to generate a transfer learning model.

III. Book Recommendation Engine using KNN

• Description: Development of a model that suggests books similar to a given book using the Nearest Neighbors algorithm.

• Files:

  • book_recommendation_knn.ipynb: Exploratory analysis of the book and rating dataframes is conducted. Model learns to identify books that are similar to each other based on the user ratings.

IV. Healthcare Cost Prediction using Regression

• Description: Predict healthcare costs of individuals using a regression algorithm.

• Files:

  • predict_health_costs_with_regression.ipynb: Preprocessing involves the transformation of both numeric and categorical features. During the modeling stage, linear regression and gradient boosting models serve as a baseline and are compared to a dense NN, employed with the premise to capture more complex relationships within the data.

V. SMS Spam Classification

• Description: Introduction to NLP, by a machine learning model that predicts whether an SMS message is "ham" or "spam."

• Files:

  • sms_text_classification.ipynb: An imbalance in labels is idintified, resulting to thoughtful consideration to determine the appropriate course of action. The text messages undergo preprocessing, and a bag-of-words model is constructed using linear regression as a baseline. Additionally, a dense neural network is implemented for comparative analysis, based on word embeddings.

Setting Up

git clone https://github.com/efthimisfytsilis/freeCodeCamp.git

cd your-local-path/freeCodeCamp

• List of dependencies to run the projects: requirements.txt.

Contributions and Feedback

I hope this repository serves as a valuable resource for exploring machine learning concepts in Python, catering to those in the early stages of their journey.

Lastly, I want to convey special thanks to the author of this Reddit post for offering such useful guidelines!