lstm_code.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "bnjmFZLExdrd"
   },
   "source": [
    "# LTFS Data Science FinHack 3 Solution"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "id": "3fG7xnCPxdro"
   },
   "outputs": [],
   "source": [
    "%reload_ext autoreload\n",
    "%autoreload 2\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "# !pip install --upgrade --no-cache-dir catboost\n",
    "# !pip install --upgrade --no-cache-dir seaborn\n",
    "# !pip install --upgrade --no-cache-dir tabulate\n",
    "# !pip install --upgrade --no-cache-dir featuretools\n",
    "# !pip install --upgrade --no-cache-dir xgboost\n",
    "# !pip install --upgrade --no-cache-dir lightgbm\n",
    "# !pip install --upgrade --no-cache-dir tqdm\n",
    "# !pip install --upgrade --no-cache-dir psutil\n",
    "# !pip install --upgrade --no-cache-dir openpyxl\n",
    "# !pip install --upgrade --no-cache-dir scikit-learn\n",
    "# !pip install --upgrade --no-cache-dir pydot2\n",
    "# !pip install --upgrade --no-cache-dir category-encoders\n",
    "# !pip install --upgrade --no-cache-dir jupyter\n",
    "# !pip install --upgrade --no-cache-dir pandas\n",
    "# !pip install --upgrade --no-cache-dir tensorflow-gpu\n",
    "# !pip install --upgrade --no-cache-dir pydot\n",
    "# !pip install --upgrade --no-cache-dir pydot2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "id": "8TeCsKpJxdrp"
   },
   "outputs": [],
   "source": [
    "import os\n",
    "import re\n",
    "import glob\n",
    "import time\n",
    "import psutil\n",
    "import joblib\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "sns.set_style(\"whitegrid\", {'axes.grid' : False})\n",
    "\n",
    "from tqdm.notebook import tqdm\n",
    "tqdm.pandas()\n",
    "\n",
    "import featuretools as ft\n",
    "\n",
    "from pathlib import Path\n",
    "from datetime import datetime\n",
    "from tabulate import tabulate\n",
    "from collections import Counter\n",
    "from collections import namedtuple\n",
    "\n",
    "from category_encoders import OrdinalEncoder\n",
    "\n",
    "import sklearn\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.utils.class_weight import compute_class_weight\n",
    "from sklearn.preprocessing import FunctionTransformer, RobustScaler, StandardScaler\n",
    "from sklearn.model_selection import KFold, StratifiedKFold, GroupKFold\n",
    "from sklearn.impute import SimpleImputer, MissingIndicator\n",
    "\n",
    "import tensorflow as tf\n",
    "import tensorflow.keras.backend as K\n",
    "from tensorflow.keras.callbacks import EarlyStopping, ModelCheckpoint, TerminateOnNaN\n",
    "from tensorflow.keras.models import Sequential, load_model, Model\n",
    "from tensorflow.keras.layers import Dense, Lambda, Flatten, Dropout, Input, Activation, Concatenate, concatenate, Bidirectional, LSTM, Convolution1D, BatchNormalization\n",
    "from tensorflow.keras.layers import Reshape, Input, Embedding, GlobalMaxPooling1D\n",
    "from tensorflow.keras.optimizers import Adam, RMSprop, SGD, Adadelta, Adagrad, Nadam\n",
    "from tensorflow.keras.utils import to_categorical, plot_model, model_to_dot\n",
    "from tensorflow.keras.regularizers import l2\n",
    "from tensorflow.keras.preprocessing.sequence import pad_sequences\n",
    "from tensorflow.keras.initializers import RandomNormal\n",
    "from tensorflow.keras.wrappers.scikit_learn import KerasClassifier\n",
    "from tensorflow.keras.losses import binary_crossentropy, categorical_crossentropy, mean_squared_error, logcosh, mean_absolute_percentage_error"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[name: \"/device:CPU:0\"\n",
      "device_type: \"CPU\"\n",
      "memory_limit: 268435456\n",
      "locality {\n",
      "}\n",
      "incarnation: 18408919460573108686\n",
      "]\n"
     ]
    }
   ],
   "source": [
    "from tensorflow.python.client import device_lib \n",
    "print(device_lib.list_local_devices())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "id": "FTR2D7d3xdrq"
   },
   "outputs": [],
   "source": [
    "def get_embedding_size(n_cat: int) -> int:\n",
    "    \"\"\"\n",
    "    Determines the embedding vector size for number of categories\n",
    "    https://github.com/fastai/fastai/blob/96c5927648ecf83f0bc9ab601f672d3c0ffe0059/fastai/tabular/data.py#L13\n",
    "    :param n_cat: number of categories\n",
    "    :return: int\n",
    "    \"\"\"\n",
    "    return int(min(600, round(1.6 * n_cat ** 0.56)))\n",
    "\n",
    "def get_embeddings(categorical_map: list):\n",
    "    \"\"\"\n",
    "    Generates embedding layers using categorical mappings\n",
    "    :param categorical_map: list, [(column_name, num_of_categories), ....]\n",
    "    :return: list, list\n",
    "    \"\"\"\n",
    "    embedding_inputs = []\n",
    "    embedding_outputs = []\n",
    "    for column_name, nb_unique_classes in categorical_map:\n",
    "        assert (nb_unique_classes > 1), f\"{column_name} column has nb_unique_classes <= 1\"\n",
    "        embedding_size = get_embedding_size(nb_unique_classes)\n",
    "\n",
    "        # One Embedding Layer for each categorical variable\n",
    "        model_input = Input(shape=(1,))\n",
    "        model_output = Embedding(nb_unique_classes, embedding_size)(model_input)\n",
    "        model_output = Reshape(target_shape=(embedding_size,))(model_output)\n",
    "\n",
    "        embedding_inputs.append(model_input)\n",
    "        embedding_outputs.append(model_output)\n",
    "\n",
    "    return embedding_inputs, embedding_outputs\n",
    "\n",
    "def get_categorical_inputs_for_dl(x_arr):\n",
    "    \"\"\"\n",
    "    https://machinelearningmastery.com/howprepare_categorical_inputs_test-to-prepare-categorical-data-for-deep-learning-in-python/\n",
    "    :param x_arr: numpy array of categorical data of shape (n_samples, n_columns)\n",
    "    :return: list\n",
    "    \"\"\"\n",
    "    return [x_arr[:, i] for i in range(x_arr.shape[1])]\n",
    "\n",
    "def plot_countplot(series_):\n",
    "    \"\"\"\n",
    "    https://stackoverflow.com/questions/46623583/seaborn-countplot-order-categories-by-count\n",
    "    :param series_:\n",
    "    :return:\n",
    "    \"\"\"\n",
    "    sns.set_style(\"whitegrid\", {'axes.grid': False})\n",
    "    headers = [series_.name, 'Count', 'Percentage']\n",
    "    value_counts = series_.value_counts()\n",
    "    value_counts_norm = series_.value_counts(normalize=True)\n",
    "    table = [[\n",
    "        v, value_counts[v], round(value_counts_norm[v], 3)\n",
    "    ] for v in value_counts.index]\n",
    "    print(tabulate(table, headers, tablefmt=\"psql\"))\n",
    "    sns.countplot(y=series_, orient='horizontal', order=value_counts.index)\n",
    "\n",
    "def date_to_integer(dt_time):\n",
    "    \"\"\"\n",
    "    https://github.com/SudalaiRajkumar/ML/blob/master/AV_LordOfTheMachines/Explorations.ipynb\n",
    "    \"\"\"\n",
    "    try:\n",
    "        return time.mktime(dt_time.timetuple())\n",
    "    except ValueError:\n",
    "        return np.nan\n",
    "\n",
    "def run_LGB(params, train, test, feature_names, n_folds=10, seed=0, return_models=False):\n",
    "    \"\"\"\n",
    "    https://stackoverflow.com/questions/56303572/how-can-i-solve-this-unknown-label-type-error\n",
    "    \"\"\"\n",
    "    skf = GroupKFold(n_splits=n_folds) # shuffle=True, random_state=seed\n",
    "    X, y, groups = train[feature_names], train['Top-up Month'].values.astype('int'), train['ID'].values.astype('int')\n",
    "\n",
    "    num_classes = np.unique(y).size\n",
    "    preds = np.zeros((test.shape[0], params['num_classes']))\n",
    "\n",
    "    models = list()\n",
    "    for i, (train_index, test_index) in enumerate(skf.split(X, y, groups)):\n",
    "        X_train, X_val = X.iloc[train_index, :], X.iloc[test_index, :]\n",
    "        y_train, y_val = y[train_index], y[test_index]\n",
    "\n",
    "        dtrain = lgb.Dataset(X_train, y_train)\n",
    "        dval = lgb.Dataset(X_val, y_val, reference=dtrain)\n",
    "\n",
    "        bst = lgb.train(\n",
    "            params=params,\n",
    "            train_set=dtrain,\n",
    "            valid_sets=[dtrain, dval],\n",
    "            valid_names=['train', 'eval'],\n",
    "            num_boost_round=30_000,\n",
    "            verbose_eval=100,\n",
    "            early_stopping_rounds=100\n",
    "        )\n",
    "\n",
    "        score_, iter_ = bst.best_score['eval']['multi_logloss'], bst.best_iteration\n",
    "        test_preds = bst.predict(test[feature_names], num_iteration=iter_)\n",
    "        preds += test_preds\n",
    "        models.append(bst)\n",
    "        \n",
    "    if return_models:\n",
    "        return preds / n_folds, models\n",
    "    return preds / n_folds\n",
    "\n",
    "def run_XGB(params, train, test, feature_names, n_folds = 10, seed = 0, return_models=False):\n",
    "    skf = GroupKFold(n_splits=n_folds)\n",
    "    X, y, groups = train[feature_names], train['Top-up Month'].values.astype('int'), train['ID'].values.astype('int')\n",
    "    \n",
    "    preds = np.zeros((test.shape[0], params['num_class']))\n",
    "    \n",
    "    models = list()\n",
    "    for i, (train_index, test_index) in enumerate(skf.split(X, y, groups)):\n",
    "        X_train, X_val = X.iloc[train_index, :], X.iloc[test_index, :]\n",
    "        y_train, y_val = y[train_index], y[test_index]\n",
    "\n",
    "        dtrain = xgb.DMatrix(data=X_train, label=y_train, missing=np.nan)\n",
    "        dval = xgb.DMatrix(data = X_val, label=y_val, missing=np.nan)\n",
    "\n",
    "        bst = xgb.train(\n",
    "                params                = params,\n",
    "                dtrain                = dtrain,\n",
    "                num_boost_round       = 30_000,\n",
    "                early_stopping_rounds = 100,\n",
    "                evals                 = [(dtrain, 'train'), (dval, 'eval')],\n",
    "                verbose_eval          = 100\n",
    "            )\n",
    "\n",
    "        score_, iter_ = bst.best_score, bst.best_iteration\n",
    "        test_preds = bst.predict(xgb.DMatrix(test[feature_names]))\n",
    "        preds += test_preds\n",
    "        models.append(bst)\n",
    "        \n",
    "    if return_models:\n",
    "        return preds / n_folds, models\n",
    "    return preds / n_folds\n",
    "\n",
    "def run_CGB(params, train, test, feature_names, n_folds = 10, seed=0, cat_cols=None, return_models=False):\n",
    "    skf = GroupKFold(n_splits=n_folds)\n",
    "    X, y, groups = train[feature_names], train['Top-up Month'].values.astype('int'), train['ID'].values.astype('int')\n",
    "    \n",
    "    preds = np.zeros((test.shape[0], params['classes_count']))\n",
    "    \n",
    "    models = list()\n",
    "    for i, (train_index, test_index) in enumerate(skf.split(X, y, groups)):\n",
    "        X_train, X_val = X.iloc[train_index, :], X.iloc[test_index, :]\n",
    "        y_train, y_val = y[train_index], y[test_index]\n",
    "\n",
    "        dtrain = cgb.Pool(data=X_train, label=y_train, cat_features=cat_cols)\n",
    "        dval = cgb.Pool(data=X_val, label=y_val, cat_features=cat_cols)\n",
    "        \n",
    "        bst = cgb.train(\n",
    "                params                = params,\n",
    "                dtrain                = dtrain,\n",
    "                num_boost_round       = 30_000,\n",
    "                early_stopping_rounds = 100,\n",
    "                evals                 = [dval],\n",
    "                verbose_eval          = 100\n",
    "            )\n",
    "\n",
    "        score_, iter_ = bst.get_best_score(), bst.get_best_iteration()\n",
    "        test_preds = bst.predict(cgb.Pool(test[feature_names], cat_features=cat_cols))\n",
    "        preds += test_preds\n",
    "        models.append(bst)\n",
    "        \n",
    "    if return_models:\n",
    "        return preds / n_folds, models\n",
    "    return preds / n_folds\n",
    "\n",
    "def get_dual_features(df, feature_tuple, op = \"ratio\"):\n",
    "    if op not in (\"ratio\", \"difference\", \"multiply\"):\n",
    "        raise ValueError(\"`op` must be one of `ratio` or `difference` or `multiply`\")\n",
    "        \n",
    "    feature_dict = {}\n",
    "    for pair in feature_tuple:\n",
    "        feature_name, f1, f2 = pair\n",
    "        \n",
    "        if np.logical_or(f1 not in df.columns, f2 not in df.columns):\n",
    "            raise ValueError(f\"either of {f1}/{f2} not found...\")\n",
    "        \n",
    "        if op == 'ratio':\n",
    "            ratio = df[f1] / df[f2] if op == 'ratio' else df[f1] - df[f2]\n",
    "        elif op == 'difference':\n",
    "            ratio = df[f1] - df[f2]\n",
    "        else:\n",
    "            ratio = df[f1] * df[f2]\n",
    "            \n",
    "        feature_dict[feature_name] = ratio.values\n",
    "    return pd.DataFrame(feature_dict)\n",
    "\n",
    "def plot_feature_importance(importance, names, model_type):\n",
    "    \"\"\"\n",
    "    https://www.analyseup.com/learn-python-for-data-science/python-random-forest-feature-importance-plot.html\n",
    "    \"\"\"\n",
    "    #Create arrays from feature importance and feature names\n",
    "    feature_importance = np.array(importance)\n",
    "    feature_names = np.array(names)\n",
    "\n",
    "    #Create a DataFrame using a Dictionary\n",
    "    data={'feature_names':feature_names,'feature_importance':feature_importance}\n",
    "    fi_df = pd.DataFrame(data)\n",
    "\n",
    "    #Sort the DataFrame in order decreasing feature importance\n",
    "    fi_df.sort_values(by=['feature_importance'], ascending=False, inplace=True)\n",
    "\n",
    "    #Define size of bar plot\n",
    "    plt.figure(figsize=(12, 30))\n",
    "    #Plot Searborn bar chart\n",
    "    sns.barplot(x=fi_df['feature_importance'][:100], y=fi_df['feature_names'][:100])\n",
    "    #Add chart labels\n",
    "    plt.title(model_type + 'FEATURE IMPORTANCE')\n",
    "    plt.xlabel('FEATURE IMPORTANCE')\n",
    "    plt.ylabel('FEATURE NAMES')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "4NBX1ZpRxdrr",
    "outputId": "f83525e7-e984-40c2-ac2d-505353dcb97c"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Using 8 CPU Cores...\n"
     ]
    }
   ],
   "source": [
    "DEFAULT_CATEGORICAL_FILL = 'NOT_AVAILABLE'\n",
    "print(f\"Using {psutil.cpu_count()} CPU Cores...\")\n",
    "\n",
    "IN_COLAB = False\n",
    "path = 'storage'\n",
    "if IN_COLAB:\n",
    "    path = '/content/drive/MyDrive/Colab Notebooks'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "id": "ZSUM-2QDX25E"
   },
   "outputs": [],
   "source": [
    "df_train_data = pd.read_pickle(f'{path}/train_Data.pkl')   \n",
    "df_train_bureau = pd.read_pickle(f'{path}/train_bureau.pkl')  \n",
    "df_test_data = pd.read_pickle(f'{path}/test_Data.pkl')  \n",
    "df_test_bureau = pd.read_pickle(f'{path}/test_bureau.pkl')  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "iKuQI5Hfxdrt",
    "outputId": "a37418c8-375c-4a42-d720-6e1b39177e0e"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(128655, 26) (560844, 25) (14745, 25) (64019, 25)\n"
     ]
    }
   ],
   "source": [
    "# (128655, 26) (560844, 25) (14745, 25) (64019, 25)\n",
    "print(df_train_data.shape, df_train_bureau.shape, df_test_data.shape, df_test_bureau.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "id": "T9VlGOCPxdrv"
   },
   "outputs": [],
   "source": [
    "df_train_bureau['source'] = 'train'\n",
    "df_test_bureau['source'] = 'test'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "id": "AWoMhvlAxdrv"
   },
   "outputs": [],
   "source": [
    "df = pd.concat([df_train_data, df_test_data], axis=0)\n",
    "df_bureau = pd.concat([df_train_bureau, df_test_bureau], axis=0)\n",
    "\n",
    "df.reset_index(drop=True, inplace=True)\n",
    "df_bureau.reset_index(drop=True, inplace=True)\n",
    "df = df_bureau.merge(df, how='left', on='ID')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(624863, 51)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "performing to_datetime...\n",
      "performing cleaning...\n",
      "filling categoricals before featuretools...\n"
     ]
    }
   ],
   "source": [
    "print(\"performing to_datetime...\")\n",
    "source_dt_cols = ['DisbursalDate', 'MaturityDAte', 'AuthDate']\n",
    "source_dt_cols_seq = ['CLOSE-DT', 'DISBURSED-DT', 'DATE-REPORTED', 'LAST-PAYMENT-DATE']\n",
    "\n",
    "for col in source_dt_cols + source_dt_cols_seq:\n",
    "    df[col] = pd.to_datetime(df[col], errors='coerce')\n",
    "df.loc[df['MaturityDAte'].isnull(), 'MaturityDAte'] = df['DisbursalDate'] + (df['MaturityDAte'] - df['DisbursalDate']).median()\n",
    "\n",
    "print(\"performing cleaning...\")    \n",
    "df.loc[df['ASSET_CLASS'].isin(['1', '01']), 'ASSET_CLASS'] = 'Standard'\n",
    "df.loc[df['ASSET_CLASS'] == '2', 'ASSET_CLASS'] = 'SubStandard'\n",
    "\n",
    "df['DISBURSED-AMT/HIGH CREDIT'] = df['DISBURSED-AMT/HIGH CREDIT'].str.replace(',', '').astype(float)\n",
    "df['CURRENT-BAL'] = df['CURRENT-BAL'].str.replace(',', '').astype(float)\n",
    "df['OVERDUE-AMT'] = df['OVERDUE-AMT'].str.replace(',', '').astype(float)\n",
    "df['CREDIT-LIMIT/SANC AMT']= df['CREDIT-LIMIT/SANC AMT'].str.replace(',', '').astype(float)\n",
    "\n",
    "df['City'] = df['City'] +'-'+ df['State']\n",
    "\n",
    "print('filling categoricals before featuretools...')\n",
    "\n",
    "source_con_cols = ['Tenure', 'AssetCost', 'AmountFinance', 'DisbursalAmount', 'LTV', 'AGE', 'MonthlyIncome']\n",
    "source_con_cols_seq = ['TENURE', 'DISBURSED-AMT/HIGH CREDIT', 'CURRENT-BAL', 'OVERDUE-AMT', 'WRITE-OFF-AMT']\n",
    "\n",
    "source_cat_cols = ['Frequency', 'InstlmentMode', 'LoanStatus', 'PaymentMode', 'BranchID', 'Area', \n",
    "                   'ManufacturerID', 'SupplierID', 'SEX', 'City', 'State', 'ZiPCODE']\n",
    "source_cat_cols_seq = ['SELF-INDICATOR', 'MATCH-TYPE', 'ACCT-TYPE', 'CONTRIBUTOR-TYPE', 'OWNERSHIP-IND', \n",
    "                       'ACCOUNT-STATUS', 'INSTALLMENT-FREQUENCY', 'ASSET_CLASS']\n",
    "\n",
    "df[source_cat_cols] = df[source_cat_cols].fillna(DEFAULT_CATEGORICAL_FILL)\n",
    "df[source_cat_cols_seq] = df[source_cat_cols_seq].fillna(DEFAULT_CATEGORICAL_FILL)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "performing regex and definitions...\n"
     ]
    }
   ],
   "source": [
    "print('performing regex and definitions...')\n",
    "def strip(value, remove_spaces=False):\n",
    "    result = re.sub(r\"\\s+\", \"\" if remove_spaces else \" \", value)\n",
    "    result = result.strip()\n",
    "    return result\n",
    "\n",
    "def installment_freq(value):\n",
    "    result = re.search(r\"[a-z]+\", str(value).lower())\n",
    "    if result:\n",
    "        return strip(result.group(0))\n",
    "    else:\n",
    "        return np.nan\n",
    "    \n",
    "def installment_amt(value):\n",
    "    result = re.search(r\"[\\d,]+\", str(value).lower())\n",
    "    if result:\n",
    "        return strip(result.group(0))\n",
    "    else:\n",
    "        return np.nan\n",
    "    \n",
    "df['p_cat_seq_INSTALLMENT_FREQ'] = df['INSTALLMENT-AMT'].map(installment_freq)\n",
    "df['INSTALLMENT_AMT_CLEAN'] = df['INSTALLMENT-AMT'].map(installment_amt).str.replace(',', '').astype(float)\n",
    "df['ACTIVE_INSTALLMENT_AMT'] = df[['ACCOUNT-STATUS', 'INSTALLMENT_AMT_CLEAN']].apply(lambda r: r[1] if r[0] == 'Active' else 0., axis=1)\n",
    "\n",
    "def active_installment_amt_monthly(r):\n",
    "    if r[1] == 'biweekly':\n",
    "        return r[0] * 2\n",
    "    elif r[1] == 'weekly':\n",
    "        return r[0] * 4\n",
    "    elif r[1] == 'quarterly':\n",
    "        return r[0] / 3   \n",
    "    elif r[1] == 'annually':\n",
    "        return r[0] / 12\n",
    "    elif r[1] == 'bimonthly':\n",
    "        return r[0] / 2\n",
    "    elif r[1] == 'semi':\n",
    "        return r[0] / 6\n",
    "    else:\n",
    "        return r[0]\n",
    "    \n",
    "df['p_con_seq_ACTIVE_MONTHLY_INSTALLMENT_AMT'] = df[['ACTIVE_INSTALLMENT_AMT', 'p_cat_seq_INSTALLMENT_FREQ']].apply(active_installment_amt_monthly, axis=1)\n",
    "df['p_con_seq_MONTHLY_INSTALLMENT_AMT'] = df[['ACTIVE_INSTALLMENT_AMT', 'p_cat_seq_INSTALLMENT_FREQ']].apply(active_installment_amt_monthly, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "hnThxo80Zqgu",
    "outputId": "73e0ec00-e85b-4ea4-a5b4-ea4cb42c5d4e"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "correcting EMI...\n",
      "extracting dual features...\n",
      "extracting more features...\n"
     ]
    }
   ],
   "source": [
    "print('correcting EMI...')\n",
    "\n",
    "def func(args):\n",
    "    total_repayment = args[0] * args[1]\n",
    "    if total_repayment < args[2]:\n",
    "        return args[0] * 10.\n",
    "    return args[0]\n",
    "\n",
    "df['EMI'] = df[['EMI', 'Tenure', 'DisbursalAmount']].apply(func, axis=1)\n",
    "\n",
    "print(\"extracting dual features...\")\n",
    "\n",
    "ratio_features_list = [\n",
    "    ('p_ra_con_DisbursalAmt_AssetCost', 'DisbursalAmount', 'AssetCost'),\n",
    "    ('p_ra_con_AssetCost_MonthlyInc', 'AssetCost', 'MonthlyIncome'),\n",
    "    ('p_ra_con_EMI_MonthlyInc', 'EMI', 'MonthlyIncome'),\n",
    "    ('p_ra_con_seq_ACTIVE_MONTHLY_INSMT_AMT_MonthlyInc', 'p_con_seq_ACTIVE_MONTHLY_INSTALLMENT_AMT', 'MonthlyIncome')\n",
    "]\n",
    "\n",
    "ratio_features = get_dual_features(df, ratio_features_list)\n",
    "\n",
    "multiply_features_list = [\n",
    "    ('p_mu_con_EMI_Tenure', 'EMI', 'Tenure'),\n",
    "]\n",
    "multiply_features = get_dual_features(df, multiply_features_list, 'multiply')\n",
    "df = pd.concat([df, ratio_features, multiply_features], axis=1)\n",
    "\n",
    "difference_features_list = [\n",
    "    ('p_di_con_AssetCost_DisbursalAmt', 'AssetCost', 'DisbursalAmount'),\n",
    "    ('p_di_con_AmtFinance_DisbursalAmt', 'AmountFinance', 'DisbursalAmount'),\n",
    "    ('p_di_con_AssetCost_AmtFinance', 'AssetCost', 'AmountFinance'),\n",
    "    ('p_di_con_Totalrepayment_AssetCost', 'p_mu_con_EMI_Tenure', 'AssetCost'),\n",
    "    ('p_di_con_seq_TT', 'Tenure', 'TENURE'),\n",
    "]\n",
    "difference_features = get_dual_features(df, difference_features_list, 'difference')\n",
    "df = pd.concat([df, difference_features], axis=1)\n",
    "\n",
    "print('extracting more features...')\n",
    "\n",
    "df['p_con_REPAYMENT_RATIO'] = df[['EMI', 'Tenure', 'DisbursalAmount']].apply(lambda args: (args[0] * args[1]) / args[2], axis=1)\n",
    "df['p_cat_seq_IS_NEW_LOAN'] = (df['DISBURSED-DT'] > df['DisbursalDate']).map(int)\n",
    "df['p_con_seq_NB_DAYS_SINCE_LTFS'] = (df['DISBURSED-DT'] - df['DisbursalDate'].shift(1)) / np.timedelta64(1, 'D')\n",
    "\n",
    "df['p_con_LOAN_TENURE'] = (df['MaturityDAte'] - df['DisbursalDate']).dt.days\n",
    "df['p_con_AGE_OF_PERSON_AT_MATURITY'] = df['AGE'] + (df['Tenure'] / 12)\n",
    "\n",
    "del ratio_features, multiply_features, difference_features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "# import decimal\n",
    "\n",
    "# # create a new context for this task\n",
    "# ctx = decimal.Context()\n",
    "# # 20 digits should be enough for everyone :D\n",
    "# ctx.prec = 108\n",
    "\n",
    "# def float_to_str(f):\n",
    "#     \"\"\"\n",
    "#     Convert the given float to a string, without resorting to scientific notation\n",
    "#     https://stackoverflow.com/questions/38847690/convert-float-to-string-in-positional-format-without-scientific-notation-and-fa\n",
    "#     \"\"\"\n",
    "#     d1 = ctx.create_decimal(repr(f))\n",
    "#     return format(d1, 'f')\n",
    "\n",
    "# def dpd_hist(args):\n",
    "#     dpd_str = str(args[0]).lower()[:18]\n",
    "    \n",
    "#     if 'e+' in dpd_str:\n",
    "#         dpd_str = float_to_str(float(dpd_str))[:18]\n",
    "\n",
    "#     lst = [float(dp) for dp in list(map(''.join, zip(*[iter(re.sub(\"xxx|ddd\", \"000\", dpd_str))]*3)))]\n",
    "#     return pd.Series(lst, dtype='float')\n",
    "\n",
    "# file_path = f'{path}/df_dpd_hst.pkl'\n",
    "# if os.path.isfile(file_path):\n",
    "#     df_dpd_hst = pd.read_pickle(file_path)\n",
    "# else:\n",
    "#     df_dpd_hst = df[['DPD - HIST']].apply(dpd_hist, axis=1)\n",
    "#     df_dpd_hst.to_pickle(file_path)\n",
    "    \n",
    "# df_dpd_hst.columns = [f\"p_con_DPD_{i}\" for i in range(df_dpd_hst.columns.__len__())]\n",
    "# df = df.merge(df_dpd_hst, left_index=True, right_index=True, how='inner', suffixes=('_suffix', None))\n",
    "# df.drop([c for c in df.columns if '_suffix' in c], axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "# def bal_hist(args):\n",
    "#     cur_bal_str = str(args[0]).lower()\n",
    "#     try:\n",
    "#         if cur_bal_str == 'nan':\n",
    "#             cur_bal_str = '0'\n",
    "\n",
    "#         lst = [float(0. if c.strip() == '' else c) / (args[1] + 1e-8) for c in cur_bal_str.rstrip(',').split(\",\")[:6]]\n",
    "#     except ValueError:\n",
    "#         print(cur_bal_str)\n",
    "#     return pd.Series(lst, dtype='float')\n",
    "\n",
    "# file_path = f'{path}/df_cur_bal_hst.pkl'\n",
    "# if os.path.isfile(file_path):\n",
    "#     df_grp = pd.read_pickle(file_path)\n",
    "# else:\n",
    "#     df_grp = df[['CUR BAL - HIST', 'DISBURSED-AMT/HIGH CREDIT']].apply(bal_hist, axis=1)\n",
    "#     df_grp.to_pickle(file_path)\n",
    "\n",
    "# df_grp.columns = [f\"p_con_CUR_BAL_{i}\" for i in range(df_grp.columns.__len__())]\n",
    "# df = df.merge(df_grp, left_index=True, right_index=True, how='inner', suffixes=('_suffix', None))\n",
    "# df.drop([c for c in df.columns if '_suffix' in c], axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "# file_path = f'{path}/df_amt_ovr_hst.pkl'\n",
    "# if os.path.isfile(file_path):\n",
    "#     df_grp = pd.read_pickle(file_path)\n",
    "# else:\n",
    "#     df_grp = df[['AMT OVERDUE - HIST', 'DISBURSED-AMT/HIGH CREDIT']].apply(bal_hist, axis=1)\n",
    "#     df_grp.to_pickle(file_path)\n",
    "\n",
    "# df_grp.columns = [f\"p_con_AMT_OVR_{i}\" for i in range(df_grp.columns.__len__())]\n",
    "# df = df.merge(df_grp, left_index=True, right_index=True, how='inner', suffixes=('_suffix', None))\n",
    "# df.drop([c for c in df.columns if '_suffix' in c], axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "# file_path = f'{path}/df_amt_paid_hst.pkl'\n",
    "# if os.path.isfile(file_path):\n",
    "#     df_grp = pd.read_pickle(file_path)\n",
    "# else:\n",
    "#     df_grp = df[['AMT PAID - HIST', 'DISBURSED-AMT/HIGH CREDIT']].apply(bal_hist, axis=1)\n",
    "#     df_grp.to_pickle(file_path)\n",
    "\n",
    "# df_grp.columns = [f\"p_con_AMT_PAID_{i}\" for i in range(df_grp.columns.__len__())]\n",
    "# df = df.merge(df_grp, left_index=True, right_index=True, how='inner', suffixes=('_suffix', None))\n",
    "# df.drop([c for c in df.columns if '_suffix' in c], axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "3EOENrJfZqgv",
    "outputId": "82b1e390-dc50-48ca-cce0-1f5f4b87f3fe"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "categorical binning...\n"
     ]
    }
   ],
   "source": [
    "print('categorical binning...')\n",
    "df['LTV_BINS'] = df['LTV'].map(int)\n",
    "df['AGE_BINS'] = df['AGE'].fillna(0).map(int)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "CznnaQ4bZqgv",
    "outputId": "c3705fcf-3e82-450e-c8cc-139805a8fad4"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "categorical pairing...\n"
     ]
    }
   ],
   "source": [
    "print('categorical pairing...')\n",
    "df['ACCT_CONTRIBUTOR_ID'] = df['ACCT-TYPE'].map(str) + df['CONTRIBUTOR-TYPE'].map(str)\n",
    "df['ACCT_OWNERSHIP_ID'] = df['ACCT-TYPE'].map(str) + df['OWNERSHIP-IND'].map(str)\n",
    "df['CONTRIBUTOR_OWNERSHIP_ID'] = df['CONTRIBUTOR-TYPE'].map(str) + df['OWNERSHIP-IND'].map(str)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "Ne2SF8GVZqgw",
    "outputId": "01ac9135-f19c-4aed-84c7-0e37b10116e7"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "binary target binning...\n"
     ]
    }
   ],
   "source": [
    "print('binary target binning...')\n",
    "df['F_NO_TOP_UP'] = df['Top-up Month'].map(lambda t: 1 if t == 'No Top-up Service' else 0)\n",
    "df['F_48_MONTHS'] = df['Top-up Month'].map(lambda t: 1 if t == ' > 48 Months' else 0)\n",
    "df['F_24_30_MONTHS'] = df['Top-up Month'].map(lambda t: 1 if t == '24-30 Months' else 0)\n",
    "df['F_12_18_MONTHS'] = df['Top-up Month'].map(lambda t: 1 if t == '12-18 Months' else 0)\n",
    "df['F_18_24_MONTHS'] = df['Top-up Month'].map(lambda t: 1 if t == '18-24 Months' else 0)\n",
    "df['F_30_36_MONTHS'] = df['Top-up Month'].map(lambda t: 1 if t == '30-36 Months' else 0)\n",
    "df['F_36_48_MONTHS'] = df['Top-up Month'].map(lambda t: 1 if t == '36-48 Months' else 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "preparing data for featuretools...\n"
     ]
    }
   ],
   "source": [
    "print('preparing data for featuretools...')\n",
    "df['REPORT_ID'] = range(df.__len__())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "# print('creating certain group features...')\n",
    "# def func(df):\n",
    "#     df = df.sort_values('DISBURSED-DT')\n",
    "    \n",
    "#     if df['DISBURSED-DT'].isnull().all():\n",
    "#         df['LOAN_RECUR_FREQ_DAYS'] = np.nan\n",
    "#     else:\n",
    "#         df['LOAN_RECUR_FREQ_DAYS'] = (df['DISBURSED-DT'] - df['DISBURSED-DT'].shift(1)) / np.timedelta64(1, 'D')\n",
    "        \n",
    "# #     df['MOST_LATEST_PAYMENT_DATE'] = df['LAST-PAYMENT-DATE'].max()\n",
    "# #     df['MOST_LATEST_DISBURSAL_DATE'] = df['DISBURSED-DT'].max()\n",
    "    \n",
    "#     return df\n",
    "\n",
    "# essential_columns = [\n",
    "#     'REPORT_ID', \n",
    "#     'ID', \n",
    "#     'DISBURSED-DT', \n",
    "#     'LAST-PAYMENT-DATE',\n",
    "# ]\n",
    "\n",
    "# jb = ['REPORT_ID']\n",
    "\n",
    "# file_path = f'{path}/df_id_groups.pkl'\n",
    "# if os.path.isfile(file_path):\n",
    "#     df_grp = pd.read_pickle(file_path)\n",
    "# else:\n",
    "#     df_grp = df[essential_columns].groupby('ID').progress_apply(func).reset_index(drop=True)\n",
    "#     df_grp.to_pickle(file_path)\n",
    "    \n",
    "# df = df.merge(df_grp.drop(list(set(essential_columns)^set(jb)), axis=1), on=jb, how='inner', suffixes=('_suffix', None))\n",
    "# df.drop([c for c in df.columns if '_suffix' in c], axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "# print('performing normalization of bureau and others....')\n",
    "# def func(df):\n",
    "#     columns_ = [\n",
    "#         ('p_con_NORM_WRT_LTFS_DISBURSED_AMT', 'DISBURSED-AMT/HIGH CREDIT'),\n",
    "#     ]\n",
    "    \n",
    "#     for cnew, c in columns_:\n",
    "#         self_value = df[df['SELF-INDICATOR'] == True][c].size \n",
    "#         self_value = df[df['SELF-INDICATOR'] == True][c].fillna(0.).iloc[0] if self_value > 0 else 0.\n",
    "#         df[cnew] = np.nan if df[c].isnull().all() or self_value == 0. else df[df['SELF-INDICATOR'] == False][c].mean() - self_value\n",
    "        \n",
    "#     return df\n",
    "\n",
    "# essential_columns = [\n",
    "#     'ID',\n",
    "#     'SELF-INDICATOR',\n",
    "#     'REPORT_ID',\n",
    "#     'ACCT-TYPE',\n",
    "#     'DISBURSED-AMT/HIGH CREDIT',\n",
    "# ]\n",
    "\n",
    "# jb = ['REPORT_ID']\n",
    "\n",
    "\n",
    "# file_path = f'{path}/df_acct_type_groups.pkl'\n",
    "# if os.path.isfile(file_path):\n",
    "#     df_grp = pd.read_pickle(file_path)\n",
    "# else:\n",
    "#     df_grp = df[essential_columns].groupby(['ID', 'ACCT-TYPE']).progress_apply(func).reset_index(drop=True)\n",
    "#     df_grp.to_pickle(file_path)\n",
    "    \n",
    "# df = df.merge(df_grp.drop(list(set(essential_columns)^set(jb)), axis=1), on=jb, how='left', suffixes=('_suffix', None))\n",
    "# df.drop([c for c in df.columns if '_suffix' in c], axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "correcting DISBURSED-AMT/HIGH CREDIT...\n"
     ]
    }
   ],
   "source": [
    "print('correcting DISBURSED-AMT/HIGH CREDIT...')\n",
    "df['DISBURSED-AMT/HIGH CREDIT'] = df[['CREDIT-LIMIT/SANC AMT', 'DISBURSED-AMT/HIGH CREDIT']].apply(lambda args: args[0] if args[0] > args[1] else args[1], axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "def func(df):\n",
    "    columns_ = [\n",
    "        ('p_con_seq_NORM_DISBURSED_AMT', 'DISBURSED-AMT/HIGH CREDIT'),\n",
    "        ('p_con_seq_NORM_INSTALLMENT_AMT', 'p_con_seq_MONTHLY_INSTALLMENT_AMT'),\n",
    "        ('p_con_seq_NORM_CURRENT_BAL', 'CURRENT-BAL'),\n",
    "        ('p_con_seq_NORM_OVERDUE_AMT', 'OVERDUE-AMT'),\n",
    "    ]\n",
    "    for cnew, c in columns_:\n",
    "        df[cnew] = np.nan if df[c].isnull().all() else df[c] - df[c].mean()\n",
    "        \n",
    "    return df\n",
    "\n",
    "essential_columns = [\n",
    "    'REPORT_ID',\n",
    "    'ACCT-TYPE',\n",
    "    'DISBURSED-AMT/HIGH CREDIT', \n",
    "    'p_con_seq_MONTHLY_INSTALLMENT_AMT', \n",
    "    'CURRENT-BAL', \n",
    "    'OVERDUE-AMT',\n",
    "]\n",
    "\n",
    "jb = ['REPORT_ID']\n",
    "\n",
    "file_path = f'{path}/df_groupby_ACCT_TYPE.pkl'\n",
    "if os.path.isfile(file_path):\n",
    "    df_grp = pd.read_pickle(file_path)\n",
    "else:\n",
    "    df_grp = df[essential_columns].groupby('ACCT-TYPE').progress_apply(func).reset_index(drop=True)\n",
    "    df_grp.to_pickle(file_path)\n",
    "    \n",
    "df = df.merge(df_grp.drop(list(set(essential_columns)^set(jb)), axis=1), on=jb, how='left', suffixes=('_suffix', None))\n",
    "df.drop([c for c in df.columns if '_suffix' in c], axis=1, inplace=True)\n",
    "\n",
    "\n",
    "def func(df):\n",
    "    if df['MonthlyIncome'].isnull().all():\n",
    "        df['p_con_NORM_MonthlyIncome'] = np.nan\n",
    "    else:\n",
    "        df['p_con_NORM_MonthlyIncome'] = df['MonthlyIncome'] - df['MonthlyIncome'].mean()\n",
    "        \n",
    "    return df\n",
    "\n",
    "essential_columns = [\n",
    "    'REPORT_ID',\n",
    "    'AGE_BINS',\n",
    "    'MonthlyIncome', \n",
    "]\n",
    "\n",
    "jb = ['REPORT_ID']\n",
    "\n",
    "\n",
    "file_path = f'{path}/df_groupby_AGE_BINS.pkl'\n",
    "if os.path.isfile(file_path):\n",
    "    df_grp = pd.read_pickle(file_path)\n",
    "else:\n",
    "    df_grp = df[essential_columns].groupby('AGE_BINS').progress_apply(func).reset_index(drop=True)\n",
    "    df_grp.to_pickle(file_path)\n",
    "    \n",
    "df = df.merge(df_grp.drop(list(set(essential_columns)^set(jb)), axis=1), on=jb, how='left', suffixes=('_suffix', None))\n",
    "df.drop([c for c in df.columns if '_suffix' in c], axis=1, inplace=True)\n",
    "\n",
    "def func(df):\n",
    "    if df['DisbursalAmount'].isnull().all():\n",
    "        df['p_con_NORM_DisbursalAmount'] = np.nan\n",
    "    else:\n",
    "        df['p_con_NORM_DisbursalAmount'] = df['DisbursalAmount'] - df['DisbursalAmount'].mean()\n",
    "        \n",
    "    return df\n",
    "\n",
    "essential_columns = [\n",
    "    'REPORT_ID',\n",
    "    'BranchID',\n",
    "    'DisbursalAmount', \n",
    "]\n",
    "\n",
    "jb = ['REPORT_ID']\n",
    "\n",
    "\n",
    "file_path = f'{path}/df_groupby_BranchID.pkl'\n",
    "if os.path.isfile(file_path):\n",
    "    df_grp = pd.read_pickle(file_path)\n",
    "else:\n",
    "    df_grp = df[essential_columns].groupby('BranchID').progress_apply(func).reset_index(drop=True)\n",
    "    df_grp.to_pickle(file_path)\n",
    "    \n",
    "df = df.merge(df_grp.drop(list(set(essential_columns)^set(jb)), axis=1), on=jb, how='left', suffixes=('_suffix', None))\n",
    "df.drop([c for c in df.columns if '_suffix' in c], axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "# other_cols = [c for c in df.columns if c not in ['ID'] + source_dt_cols + source_cat_cols]\n",
    "# df[df['ID'].isin(df['ID'].sample(1000).values)][['ID'] + source_dt_cols + source_cat_cols + other_cols].to_excel('sample.xlsx', index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "additional_seq_cols = [c for c in df.columns if '_seq_' in c]\n",
    "additional_con_seq_cols = [c for c in additional_seq_cols if '_con_' in c and c not in source_con_cols_seq]\n",
    "additional_cat_seq_cols = [c for c in additional_seq_cols if '_cat_' in c and c not in source_cat_cols_seq]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "additional_cat_cols = [c for c in df.columns if '_cat_' in c and c not in additional_seq_cols + source_cat_cols] + ['LTV_BINS', 'AGE_BINS']\n",
    "additional_con_cols = [c for c in df.columns if '_con_' in c and c not in additional_seq_cols + source_con_cols]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# print('creating temperory date ordinals...') # only for boostings algo's\n",
    "# for col in ['CLOSE-DT', 'DISBURSED-DT', 'DATE-REPORTED', 'LAST-PAYMENT-DATE']:\n",
    "#     df[f\"temp_{col}\"] = df[col].map(date_to_integer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 976
    },
    "id": "1waagksvZqgw",
    "outputId": "d8e1fb53-cb25-4099-e925-cac7b6bbbb13",
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "entering featuretools...\n"
     ]
    },
    {
     "data": {
      "image/svg+xml": [
       "<?xml version=\"1.0\" encoding=\"UTF-8\" standalone=\"no\"?>\n",
       "<!DOCTYPE svg PUBLIC \"-//W3C//DTD SVG 1.1//EN\"\n",
       " \"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd\">\n",
       "<!-- Generated by graphviz version 2.40.1 (20161225.0304)\n",
       " -->\n",
       "<!-- Title: %3 Pages: 1 -->\n",
       "<svg width=\"4852pt\" height=\"1346pt\"\n",
       " viewBox=\"0.00 0.00 4852.00 1346.00\" xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\">\n",
       "<g id=\"graph0\" class=\"graph\" transform=\"scale(1 1) rotate(0) translate(4 1342)\">\n",
       "<title>%3</title>\n",
       "<polygon fill=\"#ffffff\" stroke=\"transparent\" points=\"-4,4 -4,-1342 4848,-1342 4848,4 -4,4\"/>\n",
       "<!-- states -->\n",
       "<g id=\"node1\" class=\"node\">\n",
       "<title>states</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"4522.5,-.5 4522.5,-46.5 4654.5,-46.5 4654.5,-.5 4522.5,-.5\"/>\n",
       "<text text-anchor=\"middle\" x=\"4588.5\" y=\"-31.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">states (22 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"4522.5,-23.5 4654.5,-23.5 \"/>\n",
       "<text text-anchor=\"start\" x=\"4530.5\" y=\"-8.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">State : index</text>\n",
       "</g>\n",
       "<!-- areas -->\n",
       "<g id=\"node2\" class=\"node\">\n",
       "<title>areas</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"3825,-.5 3825,-46.5 3962,-46.5 3962,-.5 3825,-.5\"/>\n",
       "<text text-anchor=\"middle\" x=\"3893.5\" y=\"-31.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">areas (104 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"3825,-23.5 3962,-23.5 \"/>\n",
       "<text text-anchor=\"start\" x=\"3833\" y=\"-8.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">Area : index</text>\n",
       "</g>\n",
       "<!-- account_and_contributors -->\n",
       "<g id=\"node3\" class=\"node\">\n",
       "<title>account_and_contributors</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"0,-91 0,-137 285,-137 285,-91 0,-91\"/>\n",
       "<text text-anchor=\"middle\" x=\"142.5\" y=\"-121.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">account_and_contributors (240 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"0,-114 285,-114 \"/>\n",
       "<text text-anchor=\"start\" x=\"8\" y=\"-98.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ACCT_CONTRIBUTOR_ID : index</text>\n",
       "</g>\n",
       "<!-- account_and_ownerships -->\n",
       "<g id=\"node4\" class=\"node\">\n",
       "<title>account_and_ownerships</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"303.5,-91 303.5,-137 581.5,-137 581.5,-91 303.5,-91\"/>\n",
       "<text text-anchor=\"middle\" x=\"442.5\" y=\"-121.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">account_and_ownerships (143 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"303.5,-114 581.5,-114 \"/>\n",
       "<text text-anchor=\"start\" x=\"311.5\" y=\"-98.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ACCT_OWNERSHIP_ID : index</text>\n",
       "</g>\n",
       "<!-- contributors_and_ownerships -->\n",
       "<g id=\"node5\" class=\"node\">\n",
       "<title>contributors_and_ownerships</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"600,-91 600,-137 917,-137 917,-91 600,-91\"/>\n",
       "<text text-anchor=\"middle\" x=\"758.5\" y=\"-121.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">contributors_and_ownerships (43 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"600,-114 917,-114 \"/>\n",
       "<text text-anchor=\"start\" x=\"608\" y=\"-98.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">CONTRIBUTOR_OWNERSHIP_ID : index</text>\n",
       "</g>\n",
       "<!-- agreements -->\n",
       "<g id=\"node6\" class=\"node\">\n",
       "<title>agreements</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"935,-91 935,-137 1144,-137 1144,-91 935,-91\"/>\n",
       "<text text-anchor=\"middle\" x=\"1039.5\" y=\"-121.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">agreements (143400 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"935,-114 1144,-114 \"/>\n",
       "<text text-anchor=\"start\" x=\"943\" y=\"-98.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ID : index</text>\n",
       "</g>\n",
       "<!-- account_types -->\n",
       "<g id=\"node7\" class=\"node\">\n",
       "<title>account_types</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"1162.5,-91 1162.5,-137 1354.5,-137 1354.5,-91 1162.5,-91\"/>\n",
       "<text text-anchor=\"middle\" x=\"1258.5\" y=\"-121.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">account_types (51 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"1162.5,-114 1354.5,-114 \"/>\n",
       "<text text-anchor=\"start\" x=\"1170.5\" y=\"-98.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ACCT&#45;TYPE : index</text>\n",
       "</g>\n",
       "<!-- self_indicators -->\n",
       "<g id=\"node8\" class=\"node\">\n",
       "<title>self_indicators</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"1372.5,-91 1372.5,-137 1574.5,-137 1574.5,-91 1372.5,-91\"/>\n",
       "<text text-anchor=\"middle\" x=\"1473.5\" y=\"-121.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">self_indicators (2 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"1372.5,-114 1574.5,-114 \"/>\n",
       "<text text-anchor=\"start\" x=\"1380.5\" y=\"-98.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">SELF&#45;INDICATOR : index</text>\n",
       "</g>\n",
       "<!-- match_types -->\n",
       "<g id=\"node9\" class=\"node\">\n",
       "<title>match_types</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"1593,-91 1593,-137 1764,-137 1764,-91 1593,-91\"/>\n",
       "<text text-anchor=\"middle\" x=\"1678.5\" y=\"-121.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">match_types (2 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"1593,-114 1764,-114 \"/>\n",
       "<text text-anchor=\"start\" x=\"1601\" y=\"-98.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">MATCH&#45;TYPE : index</text>\n",
       "</g>\n",
       "<!-- contributor_types -->\n",
       "<g id=\"node10\" class=\"node\">\n",
       "<title>contributor_types</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"1782.5,-91 1782.5,-137 2012.5,-137 2012.5,-91 1782.5,-91\"/>\n",
       "<text text-anchor=\"middle\" x=\"1897.5\" y=\"-121.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">contributor_types (12 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"1782.5,-114 2012.5,-114 \"/>\n",
       "<text text-anchor=\"start\" x=\"1790.5\" y=\"-98.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">CONTRIBUTOR&#45;TYPE : index</text>\n",
       "</g>\n",
       "<!-- ownerships -->\n",
       "<g id=\"node11\" class=\"node\">\n",
       "<title>ownerships</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"2030.5,-91 2030.5,-137 2232.5,-137 2232.5,-91 2030.5,-91\"/>\n",
       "<text text-anchor=\"middle\" x=\"2131.5\" y=\"-121.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ownerships (5 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"2030.5,-114 2232.5,-114 \"/>\n",
       "<text text-anchor=\"start\" x=\"2038.5\" y=\"-98.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">OWNERSHIP&#45;IND : index</text>\n",
       "</g>\n",
       "<!-- account_statuses -->\n",
       "<g id=\"node12\" class=\"node\">\n",
       "<title>account_statuses</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"2251,-91 2251,-137 2464,-137 2464,-91 2251,-91\"/>\n",
       "<text text-anchor=\"middle\" x=\"2357.5\" y=\"-121.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">account_statuses (11 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"2251,-114 2464,-114 \"/>\n",
       "<text text-anchor=\"start\" x=\"2259\" y=\"-98.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ACCOUNT&#45;STATUS : index</text>\n",
       "</g>\n",
       "<!-- asset_classes -->\n",
       "<g id=\"node13\" class=\"node\">\n",
       "<title>asset_classes</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"2481.5,-91 2481.5,-137 2659.5,-137 2659.5,-91 2481.5,-91\"/>\n",
       "<text text-anchor=\"middle\" x=\"2570.5\" y=\"-121.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">asset_classes (6 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"2481.5,-114 2659.5,-114 \"/>\n",
       "<text text-anchor=\"start\" x=\"2489.5\" y=\"-98.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ASSET_CLASS : index</text>\n",
       "</g>\n",
       "<!-- age -->\n",
       "<g id=\"node14\" class=\"node\">\n",
       "<title>age</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"2677.5,-91 2677.5,-137 2825.5,-137 2825.5,-91 2677.5,-91\"/>\n",
       "<text text-anchor=\"middle\" x=\"2751.5\" y=\"-121.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">age (74 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"2677.5,-114 2825.5,-114 \"/>\n",
       "<text text-anchor=\"start\" x=\"2685.5\" y=\"-98.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">AGE_BINS : index</text>\n",
       "</g>\n",
       "<!-- ltv -->\n",
       "<g id=\"node15\" class=\"node\">\n",
       "<title>ltv</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"2843.5,-91 2843.5,-137 2987.5,-137 2987.5,-91 2843.5,-91\"/>\n",
       "<text text-anchor=\"middle\" x=\"2915.5\" y=\"-121.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ltv (99 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"2843.5,-114 2987.5,-114 \"/>\n",
       "<text text-anchor=\"start\" x=\"2851.5\" y=\"-98.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">LTV_BINS : index</text>\n",
       "</g>\n",
       "<!-- frequency -->\n",
       "<g id=\"node16\" class=\"node\">\n",
       "<title>frequency</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"3005.5,-91 3005.5,-137 3157.5,-137 3157.5,-91 3005.5,-91\"/>\n",
       "<text text-anchor=\"middle\" x=\"3081.5\" y=\"-121.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">frequency (4 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"3005.5,-114 3157.5,-114 \"/>\n",
       "<text text-anchor=\"start\" x=\"3013.5\" y=\"-98.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">Frequency : index</text>\n",
       "</g>\n",
       "<!-- installment_modes -->\n",
       "<g id=\"node17\" class=\"node\">\n",
       "<title>installment_modes</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"3176,-91 3176,-137 3391,-137 3391,-91 3176,-91\"/>\n",
       "<text text-anchor=\"middle\" x=\"3283.5\" y=\"-121.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">installment_modes (2 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"3176,-114 3391,-114 \"/>\n",
       "<text text-anchor=\"start\" x=\"3184\" y=\"-98.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">InstlmentMode : index</text>\n",
       "</g>\n",
       "<!-- loan_status -->\n",
       "<g id=\"node18\" class=\"node\">\n",
       "<title>loan_status</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"3409,-91 3409,-137 3570,-137 3570,-91 3409,-91\"/>\n",
       "<text text-anchor=\"middle\" x=\"3489.5\" y=\"-121.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">loan_status (2 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"3409,-114 3570,-114 \"/>\n",
       "<text text-anchor=\"start\" x=\"3417\" y=\"-98.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">LoanStatus : index</text>\n",
       "</g>\n",
       "<!-- payment_modes -->\n",
       "<g id=\"node19\" class=\"node\">\n",
       "<title>payment_modes</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"3588,-91 3588,-137 3793,-137 3793,-91 3588,-91\"/>\n",
       "<text text-anchor=\"middle\" x=\"3690.5\" y=\"-121.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">payment_modes (11 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"3588,-114 3793,-114 \"/>\n",
       "<text text-anchor=\"start\" x=\"3596\" y=\"-98.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">PaymentMode : index</text>\n",
       "</g>\n",
       "<!-- branches -->\n",
       "<g id=\"node20\" class=\"node\">\n",
       "<title>branches</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"3811.5,-83.5 3811.5,-144.5 3975.5,-144.5 3975.5,-83.5 3811.5,-83.5\"/>\n",
       "<text text-anchor=\"middle\" x=\"3893.5\" y=\"-129.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">branches (214 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"3811.5,-121.5 3975.5,-121.5 \"/>\n",
       "<text text-anchor=\"start\" x=\"3819.5\" y=\"-106.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">BranchID : index</text>\n",
       "<text text-anchor=\"start\" x=\"3819.5\" y=\"-91.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">Area : id</text>\n",
       "</g>\n",
       "<!-- branches&#45;&gt;areas -->\n",
       "<g id=\"edge25\" class=\"edge\">\n",
       "<title>branches&#45;&gt;areas</title>\n",
       "<path fill=\"none\" stroke=\"#000000\" d=\"M3893.5,-83.3608C3893.5,-83.3608 3893.5,-56.6665 3893.5,-56.6665\"/>\n",
       "<polygon fill=\"#000000\" stroke=\"#000000\" points=\"3897.0001,-56.6664 3893.5,-46.6665 3890.0001,-56.6665 3897.0001,-56.6664\"/>\n",
       "<text text-anchor=\"middle\" x=\"3876.5\" y=\"-58.8136\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">Area</text>\n",
       "</g>\n",
       "<!-- suppliers -->\n",
       "<g id=\"node21\" class=\"node\">\n",
       "<title>suppliers</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"3994,-91 3994,-137 4167,-137 4167,-91 3994,-91\"/>\n",
       "<text text-anchor=\"middle\" x=\"4080.5\" y=\"-121.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">suppliers (4971 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"3994,-114 4167,-114 \"/>\n",
       "<text text-anchor=\"start\" x=\"4002\" y=\"-98.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">SupplierID : index</text>\n",
       "</g>\n",
       "<!-- manufacturers -->\n",
       "<g id=\"node22\" class=\"node\">\n",
       "<title>manufacturers</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"4185.5,-91 4185.5,-137 4379.5,-137 4379.5,-91 4185.5,-91\"/>\n",
       "<text text-anchor=\"middle\" x=\"4282.5\" y=\"-121.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">manufacturers (14 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"4185.5,-114 4379.5,-114 \"/>\n",
       "<text text-anchor=\"start\" x=\"4193.5\" y=\"-98.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ManufacturerID : index</text>\n",
       "</g>\n",
       "<!-- sex -->\n",
       "<g id=\"node23\" class=\"node\">\n",
       "<title>sex</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"4397.5,-91 4397.5,-137 4501.5,-137 4501.5,-91 4397.5,-91\"/>\n",
       "<text text-anchor=\"middle\" x=\"4449.5\" y=\"-121.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">sex (3 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"4397.5,-114 4501.5,-114 \"/>\n",
       "<text text-anchor=\"start\" x=\"4405.5\" y=\"-98.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">SEX : index</text>\n",
       "</g>\n",
       "<!-- cities -->\n",
       "<g id=\"node24\" class=\"node\">\n",
       "<title>cities</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"4520,-83.5 4520,-144.5 4657,-144.5 4657,-83.5 4520,-83.5\"/>\n",
       "<text text-anchor=\"middle\" x=\"4588.5\" y=\"-129.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">cities (536 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"4520,-121.5 4657,-121.5 \"/>\n",
       "<text text-anchor=\"start\" x=\"4528\" y=\"-106.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">City : index</text>\n",
       "<text text-anchor=\"start\" x=\"4528\" y=\"-91.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">State : id</text>\n",
       "</g>\n",
       "<!-- cities&#45;&gt;states -->\n",
       "<g id=\"edge24\" class=\"edge\">\n",
       "<title>cities&#45;&gt;states</title>\n",
       "<path fill=\"none\" stroke=\"#000000\" d=\"M4588.5,-83.3608C4588.5,-83.3608 4588.5,-56.6665 4588.5,-56.6665\"/>\n",
       "<polygon fill=\"#000000\" stroke=\"#000000\" points=\"4592.0001,-56.6664 4588.5,-46.6665 4585.0001,-56.6665 4592.0001,-56.6664\"/>\n",
       "<text text-anchor=\"middle\" x=\"4569\" y=\"-58.8136\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">State</text>\n",
       "</g>\n",
       "<!-- zipcodes -->\n",
       "<g id=\"node25\" class=\"node\">\n",
       "<title>zipcodes</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"4675,-91 4675,-137 4844,-137 4844,-91 4675,-91\"/>\n",
       "<text text-anchor=\"middle\" x=\"4759.5\" y=\"-121.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">zipcodes (9909 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"4675,-114 4844,-114 \"/>\n",
       "<text text-anchor=\"start\" x=\"4683\" y=\"-98.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ZiPCODE : index</text>\n",
       "</g>\n",
       "<!-- reports -->\n",
       "<g id=\"node26\" class=\"node\">\n",
       "<title>reports</title>\n",
       "<polygon fill=\"none\" stroke=\"#000000\" points=\"2485.5,-181.5 2485.5,-1337.5 3017.5,-1337.5 3017.5,-181.5 2485.5,-181.5\"/>\n",
       "<text text-anchor=\"middle\" x=\"2751.5\" y=\"-1322.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">reports (624863 rows)</text>\n",
       "<polyline fill=\"none\" stroke=\"#000000\" points=\"2485.5,-1314.5 3017.5,-1314.5 \"/>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-1299.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">REPORT_ID : index</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-1284.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">AuthDate : datetime</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-1269.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">MaturityDAte : datetime</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-1254.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">DisbursalDate : datetime</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-1239.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">CLOSE&#45;DT : datetime</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-1224.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">DISBURSED&#45;DT : datetime</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-1209.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">DATE&#45;REPORTED : datetime</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-1194.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">LAST&#45;PAYMENT&#45;DATE : datetime</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-1179.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ACCT_CONTRIBUTOR_ID : id</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-1164.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ACCT_OWNERSHIP_ID : id</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-1149.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">CONTRIBUTOR_OWNERSHIP_ID : id</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-1134.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ID : id</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-1119.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ACCT&#45;TYPE : id</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-1104.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">SELF&#45;INDICATOR : id</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-1089.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">MATCH&#45;TYPE : id</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-1074.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">CONTRIBUTOR&#45;TYPE : id</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-1059.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">OWNERSHIP&#45;IND : id</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-1044.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ACCOUNT&#45;STATUS : id</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-1029.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ASSET_CLASS : id</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-1014.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">AGE_BINS : id</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-999.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">LTV_BINS : id</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-984.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">Frequency : id</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-969.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">InstlmentMode : id</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-954.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">LoanStatus : id</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-939.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">PaymentMode : id</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-924.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">BranchID : id</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-909.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">SupplierID : id</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-894.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ManufacturerID : id</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-879.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">SEX : id</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-864.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">City : id</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-849.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ZiPCODE : id</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-834.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">Tenure : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-819.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">AssetCost : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-804.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">AmountFinance : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-789.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">DisbursalAmount : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-774.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">EMI : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-759.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">LTV : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-744.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">AGE : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-729.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">MonthlyIncome : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-714.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">p_ra_con_DisbursalAmt_AssetCost : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-699.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">p_ra_con_AssetCost_MonthlyInc : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-684.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">p_ra_con_EMI_MonthlyInc : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-669.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">p_mu_con_EMI_Tenure : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-654.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">p_di_con_AssetCost_DisbursalAmt : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-639.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">p_di_con_AmtFinance_DisbursalAmt : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-624.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">p_di_con_AssetCost_AmtFinance : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-609.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">p_di_con_Totalrepayment_AssetCost : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-594.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">p_con_REPAYMENT_RATIO : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-579.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">p_con_LOAN_TENURE : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-564.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">p_con_AGE_OF_PERSON_AT_MATURITY : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-549.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">p_con_NORM_MonthlyIncome : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-534.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">p_con_NORM_DisbursalAmount : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-519.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">TENURE : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-504.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">DISBURSED&#45;AMT/HIGH CREDIT : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-489.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">CURRENT&#45;BAL : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-474.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">OVERDUE&#45;AMT : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-459.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">WRITE&#45;OFF&#45;AMT : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-444.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">p_con_seq_ACTIVE_MONTHLY_INSTALLMENT_AMT : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-429.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">p_con_seq_MONTHLY_INSTALLMENT_AMT : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-414.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">p_ra_con_seq_ACTIVE_MONTHLY_INSMT_AMT_MonthlyInc : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-399.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">p_di_con_seq_TT : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-384.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">p_con_seq_NB_DAYS_SINCE_LTFS : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-369.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">p_con_seq_NORM_DISBURSED_AMT : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-354.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">p_con_seq_NORM_INSTALLMENT_AMT : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-339.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">p_con_seq_NORM_CURRENT_BAL : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-324.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">p_con_seq_NORM_OVERDUE_AMT : numeric</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-309.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">p_cat_seq_INSTALLMENT_FREQ : categorical</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-294.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">p_cat_seq_IS_NEW_LOAN : categorical</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-279.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">F_NO_TOP_UP : boolean</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-264.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">F_48_MONTHS : boolean</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-249.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">F_24_30_MONTHS : boolean</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-234.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">F_12_18_MONTHS : boolean</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-219.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">F_18_24_MONTHS : boolean</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-204.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">F_30_36_MONTHS : boolean</text>\n",
       "<text text-anchor=\"start\" x=\"2493.5\" y=\"-189.3\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">F_36_48_MONTHS : boolean</text>\n",
       "</g>\n",
       "<!-- reports&#45;&gt;account_and_contributors -->\n",
       "<g id=\"edge1\" class=\"edge\">\n",
       "<title>reports&#45;&gt;account_and_contributors</title>\n",
       "<path fill=\"none\" stroke=\"#000000\" d=\"M2485.4627,-1232C1821.1383,-1232 142.5,-1232 142.5,-1232 142.5,-1232 142.5,-147.0944 142.5,-147.0944\"/>\n",
       "<polygon fill=\"#000000\" stroke=\"#000000\" points=\"146.0001,-147.0944 142.5,-137.0944 139.0001,-147.0944 146.0001,-147.0944\"/>\n",
       "<text text-anchor=\"middle\" x=\"675.0285\" y=\"-1235.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ACCT_CONTRIBUTOR_ID</text>\n",
       "</g>\n",
       "<!-- reports&#45;&gt;account_and_ownerships -->\n",
       "<g id=\"edge2\" class=\"edge\">\n",
       "<title>reports&#45;&gt;account_and_ownerships</title>\n",
       "<path fill=\"none\" stroke=\"#000000\" d=\"M2485.2194,-1127C1877.9055,-1127 442.5,-1127 442.5,-1127 442.5,-1127 442.5,-147.0612 442.5,-147.0612\"/>\n",
       "<polygon fill=\"#000000\" stroke=\"#000000\" points=\"446.0001,-147.0612 442.5,-137.0612 439.0001,-147.0613 446.0001,-147.0612\"/>\n",
       "<text text-anchor=\"middle\" x=\"886.3903\" y=\"-1130.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ACCT_OWNERSHIP_ID</text>\n",
       "</g>\n",
       "<!-- reports&#45;&gt;contributors_and_ownerships -->\n",
       "<g id=\"edge3\" class=\"edge\">\n",
       "<title>reports&#45;&gt;contributors_and_ownerships</title>\n",
       "<path fill=\"none\" stroke=\"#000000\" d=\"M2485.2992,-1022C1942.4587,-1022 758.5,-1022 758.5,-1022 758.5,-1022 758.5,-147.3575 758.5,-147.3575\"/>\n",
       "<polygon fill=\"#000000\" stroke=\"#000000\" points=\"762.0001,-147.3575 758.5,-137.3575 755.0001,-147.3576 762.0001,-147.3575\"/>\n",
       "<text text-anchor=\"middle\" x=\"1060.0784\" y=\"-1025.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">CONTRIBUTOR_OWNERSHIP_ID</text>\n",
       "</g>\n",
       "<!-- reports&#45;&gt;agreements -->\n",
       "<g id=\"edge4\" class=\"edge\">\n",
       "<title>reports&#45;&gt;agreements</title>\n",
       "<path fill=\"none\" stroke=\"#000000\" d=\"M2485.4093,-917C2004.3818,-917 1039.5,-917 1039.5,-917 1039.5,-917 1039.5,-147.1662 1039.5,-147.1662\"/>\n",
       "<polygon fill=\"#000000\" stroke=\"#000000\" points=\"1043.0001,-147.1662 1039.5,-137.1662 1036.0001,-147.1663 1043.0001,-147.1662\"/>\n",
       "<text text-anchor=\"middle\" x=\"1368.5377\" y=\"-920.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ID</text>\n",
       "</g>\n",
       "<!-- reports&#45;&gt;account_types -->\n",
       "<g id=\"edge5\" class=\"edge\">\n",
       "<title>reports&#45;&gt;account_types</title>\n",
       "<path fill=\"none\" stroke=\"#000000\" d=\"M2485.4422,-812C2056.2072,-812 1258.5,-812 1258.5,-812 1258.5,-812 1258.5,-147.2237 1258.5,-147.2237\"/>\n",
       "<polygon fill=\"#000000\" stroke=\"#000000\" points=\"1262.0001,-147.2237 1258.5,-137.2237 1255.0001,-147.2237 1262.0001,-147.2237\"/>\n",
       "<text text-anchor=\"middle\" x=\"1497.083\" y=\"-815.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ACCT&#45;TYPE</text>\n",
       "</g>\n",
       "<!-- reports&#45;&gt;self_indicators -->\n",
       "<g id=\"edge6\" class=\"edge\">\n",
       "<title>reports&#45;&gt;self_indicators</title>\n",
       "<path fill=\"none\" stroke=\"#000000\" d=\"M2485.145,-706C2110.5999,-706 1473.5,-706 1473.5,-706 1473.5,-706 1473.5,-147.2071 1473.5,-147.2071\"/>\n",
       "<polygon fill=\"#000000\" stroke=\"#000000\" points=\"1477.0001,-147.2071 1473.5,-137.2071 1470.0001,-147.2071 1477.0001,-147.2071\"/>\n",
       "<text text-anchor=\"middle\" x=\"1633.426\" y=\"-709.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">SELF&#45;INDICATOR</text>\n",
       "</g>\n",
       "<!-- reports&#45;&gt;match_types -->\n",
       "<g id=\"edge7\" class=\"edge\">\n",
       "<title>reports&#45;&gt;match_types</title>\n",
       "<path fill=\"none\" stroke=\"#000000\" d=\"M2485.2302,-601C2167.4564,-601 1678.5,-601 1678.5,-601 1678.5,-601 1678.5,-147.1614 1678.5,-147.1614\"/>\n",
       "<polygon fill=\"#000000\" stroke=\"#000000\" points=\"1682.0001,-147.1614 1678.5,-137.1614 1675.0001,-147.1614 1682.0001,-147.1614\"/>\n",
       "<text text-anchor=\"middle\" x=\"1803.9458\" y=\"-604.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">MATCH&#45;TYPE</text>\n",
       "</g>\n",
       "<!-- reports&#45;&gt;contributor_types -->\n",
       "<g id=\"edge8\" class=\"edge\">\n",
       "<title>reports&#45;&gt;contributor_types</title>\n",
       "<path fill=\"none\" stroke=\"#000000\" d=\"M2485.3863,-496C2234.7524,-496 1897.5,-496 1897.5,-496 1897.5,-496 1897.5,-147.0355 1897.5,-147.0355\"/>\n",
       "<polygon fill=\"#000000\" stroke=\"#000000\" points=\"1901.0001,-147.0355 1897.5,-137.0355 1894.0001,-147.0356 1901.0001,-147.0355\"/>\n",
       "<text text-anchor=\"middle\" x=\"1936.4609\" y=\"-499.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">CONTRIBUTOR&#45;TYPE</text>\n",
       "</g>\n",
       "<!-- reports&#45;&gt;ownerships -->\n",
       "<g id=\"edge9\" class=\"edge\">\n",
       "<title>reports&#45;&gt;ownerships</title>\n",
       "<path fill=\"none\" stroke=\"#000000\" d=\"M2485.4161,-391C2317.2561,-391 2131.5,-391 2131.5,-391 2131.5,-391 2131.5,-147.1404 2131.5,-147.1404\"/>\n",
       "<polygon fill=\"#000000\" stroke=\"#000000\" points=\"2135.0001,-147.1404 2131.5,-137.1404 2128.0001,-147.1405 2135.0001,-147.1404\"/>\n",
       "<text text-anchor=\"middle\" x=\"2120.0283\" y=\"-394.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">OWNERSHIP&#45;IND</text>\n",
       "</g>\n",
       "<!-- reports&#45;&gt;account_statuses -->\n",
       "<g id=\"edge10\" class=\"edge\">\n",
       "<title>reports&#45;&gt;account_statuses</title>\n",
       "<path fill=\"none\" stroke=\"#000000\" d=\"M2485.4205,-286C2414.6512,-286 2357.5,-286 2357.5,-286 2357.5,-286 2357.5,-147.1967 2357.5,-147.1967\"/>\n",
       "<polygon fill=\"#000000\" stroke=\"#000000\" points=\"2361.0001,-147.1967 2357.5,-137.1967 2354.0001,-147.1967 2361.0001,-147.1967\"/>\n",
       "<text text-anchor=\"middle\" x=\"2286\" y=\"-284.3586\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ACCOUNT&#45;STATUS</text>\n",
       "</g>\n",
       "<!-- reports&#45;&gt;asset_classes -->\n",
       "<g id=\"edge11\" class=\"edge\">\n",
       "<title>reports&#45;&gt;asset_classes</title>\n",
       "<path fill=\"none\" stroke=\"#000000\" d=\"M2572.5,-181.2531C2572.5,-181.2531 2572.5,-147.0167 2572.5,-147.0167\"/>\n",
       "<polygon fill=\"#000000\" stroke=\"#000000\" points=\"2576.0001,-147.0167 2572.5,-137.0167 2569.0001,-147.0168 2576.0001,-147.0167\"/>\n",
       "<text text-anchor=\"middle\" x=\"2518\" y=\"-167.9349\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ASSET_CLASS</text>\n",
       "</g>\n",
       "<!-- reports&#45;&gt;age -->\n",
       "<g id=\"edge12\" class=\"edge\">\n",
       "<title>reports&#45;&gt;age</title>\n",
       "<path fill=\"none\" stroke=\"#000000\" d=\"M2751.5,-181.2531C2751.5,-181.2531 2751.5,-147.0167 2751.5,-147.0167\"/>\n",
       "<polygon fill=\"#000000\" stroke=\"#000000\" points=\"2755.0001,-147.0167 2751.5,-137.0167 2748.0001,-147.0168 2755.0001,-147.0167\"/>\n",
       "<text text-anchor=\"middle\" x=\"2712\" y=\"-167.9349\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">AGE_BINS</text>\n",
       "</g>\n",
       "<!-- reports&#45;&gt;ltv -->\n",
       "<g id=\"edge13\" class=\"edge\">\n",
       "<title>reports&#45;&gt;ltv</title>\n",
       "<path fill=\"none\" stroke=\"#000000\" d=\"M2915.5,-181.2531C2915.5,-181.2531 2915.5,-147.0167 2915.5,-147.0167\"/>\n",
       "<polygon fill=\"#000000\" stroke=\"#000000\" points=\"2919.0001,-147.0167 2915.5,-137.0167 2912.0001,-147.0168 2919.0001,-147.0167\"/>\n",
       "<text text-anchor=\"middle\" x=\"2877.5\" y=\"-167.9349\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">LTV_BINS</text>\n",
       "</g>\n",
       "<!-- reports&#45;&gt;frequency -->\n",
       "<g id=\"edge14\" class=\"edge\">\n",
       "<title>reports&#45;&gt;frequency</title>\n",
       "<path fill=\"none\" stroke=\"#000000\" d=\"M3007.5,-181.2531C3007.5,-181.2531 3007.5,-147.0167 3007.5,-147.0167\"/>\n",
       "<polygon fill=\"#000000\" stroke=\"#000000\" points=\"3011.0001,-147.0167 3007.5,-137.0167 3004.0001,-147.0168 3011.0001,-147.0167\"/>\n",
       "<text text-anchor=\"middle\" x=\"2969\" y=\"-167.9349\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">Frequency</text>\n",
       "</g>\n",
       "<!-- reports&#45;&gt;installment_modes -->\n",
       "<g id=\"edge15\" class=\"edge\">\n",
       "<title>reports&#45;&gt;installment_modes</title>\n",
       "<path fill=\"none\" stroke=\"#000000\" d=\"M3011.5,-181.3921C3011.5,-160.9058 3011.5,-149 3011.5,-149 3011.5,-149 3283.5,-149 3283.5,-149 3283.5,-149 3283.5,-147.1506 3283.5,-147.1506\"/>\n",
       "<polygon fill=\"#000000\" stroke=\"#000000\" points=\"3287.0001,-147.1506 3283.5,-137.1506 3280.0001,-147.1506 3287.0001,-147.1506\"/>\n",
       "<text text-anchor=\"middle\" x=\"3077.2287\" y=\"-152.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">InstlmentMode</text>\n",
       "</g>\n",
       "<!-- reports&#45;&gt;loan_status -->\n",
       "<g id=\"edge16\" class=\"edge\">\n",
       "<title>reports&#45;&gt;loan_status</title>\n",
       "<path fill=\"none\" stroke=\"#000000\" d=\"M3013.5,-181.4393C3013.5,-163.3858 3013.5,-153 3013.5,-153 3013.5,-153 3489.5,-153 3489.5,-153 3489.5,-153 3489.5,-147.1372 3489.5,-147.1372\"/>\n",
       "<polygon fill=\"#000000\" stroke=\"#000000\" points=\"3493.0001,-147.1372 3489.5,-137.1372 3486.0001,-147.1372 3493.0001,-147.1372\"/>\n",
       "<text text-anchor=\"middle\" x=\"3199.2118\" y=\"-156.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">LoanStatus</text>\n",
       "</g>\n",
       "<!-- reports&#45;&gt;payment_modes -->\n",
       "<g id=\"edge17\" class=\"edge\">\n",
       "<title>reports&#45;&gt;payment_modes</title>\n",
       "<path fill=\"none\" stroke=\"#000000\" d=\"M3015.5,-181.3649C3015.5,-165.8349 3015.5,-157 3015.5,-157 3015.5,-157 3690.5,-157 3690.5,-157 3690.5,-157 3690.5,-147.0106 3690.5,-147.0106\"/>\n",
       "<polygon fill=\"#000000\" stroke=\"#000000\" points=\"3694.0001,-147.0106 3690.5,-137.0106 3687.0001,-147.0106 3694.0001,-147.0106\"/>\n",
       "<text text-anchor=\"middle\" x=\"3294.3122\" y=\"-160.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">PaymentMode</text>\n",
       "</g>\n",
       "<!-- reports&#45;&gt;branches -->\n",
       "<g id=\"edge18\" class=\"edge\">\n",
       "<title>reports&#45;&gt;branches</title>\n",
       "<path fill=\"none\" stroke=\"#000000\" d=\"M3009.5,-181.105C3009.5,-155.9119 3009.5,-141 3009.5,-141 3009.5,-141 3801.1292,-141 3801.1292,-141\"/>\n",
       "<polygon fill=\"#000000\" stroke=\"#000000\" points=\"3801.1293,-144.5001 3811.1292,-141 3801.1292,-137.5001 3801.1293,-144.5001\"/>\n",
       "<text text-anchor=\"middle\" x=\"3350.2621\" y=\"-144.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">BranchID</text>\n",
       "</g>\n",
       "<!-- reports&#45;&gt;suppliers -->\n",
       "<g id=\"edge19\" class=\"edge\">\n",
       "<title>reports&#45;&gt;suppliers</title>\n",
       "<path fill=\"none\" stroke=\"#000000\" d=\"M2990.5,-181.2561C2990.5,-168.2882 2990.5,-161 2990.5,-161 2990.5,-161 4080.5,-161 4080.5,-161 4080.5,-161 4080.5,-147.2246 4080.5,-147.2246\"/>\n",
       "<polygon fill=\"#000000\" stroke=\"#000000\" points=\"4084.0001,-147.2246 4080.5,-137.2246 4077.0001,-147.2246 4084.0001,-147.2246\"/>\n",
       "<text text-anchor=\"middle\" x=\"3492.7596\" y=\"-164.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">SupplierID</text>\n",
       "</g>\n",
       "<!-- reports&#45;&gt;manufacturers -->\n",
       "<g id=\"edge20\" class=\"edge\">\n",
       "<title>reports&#45;&gt;manufacturers</title>\n",
       "<path fill=\"none\" stroke=\"#000000\" d=\"M2993.5,-181.2098C2993.5,-170.7836 2993.5,-165 2993.5,-165 2993.5,-165 4282.5,-165 4282.5,-165 4282.5,-165 4282.5,-147.1125 4282.5,-147.1125\"/>\n",
       "<polygon fill=\"#000000\" stroke=\"#000000\" points=\"4286.0001,-147.1125 4282.5,-137.1125 4279.0001,-147.1125 4286.0001,-147.1125\"/>\n",
       "<text text-anchor=\"middle\" x=\"3697.3389\" y=\"-153.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ManufacturerID</text>\n",
       "</g>\n",
       "<!-- reports&#45;&gt;sex -->\n",
       "<g id=\"edge21\" class=\"edge\">\n",
       "<title>reports&#45;&gt;sex</title>\n",
       "<path fill=\"none\" stroke=\"#000000\" d=\"M2996.5,-181.3334C2996.5,-173.361 2996.5,-169 2996.5,-169 2996.5,-169 4449.5,-169 4449.5,-169 4449.5,-169 4449.5,-147.0063 4449.5,-147.0063\"/>\n",
       "<polygon fill=\"#000000\" stroke=\"#000000\" points=\"4453.0001,-147.0062 4449.5,-137.0063 4446.0001,-147.0063 4453.0001,-147.0062\"/>\n",
       "<text text-anchor=\"middle\" x=\"3712.3301\" y=\"-172.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">SEX</text>\n",
       "</g>\n",
       "<!-- reports&#45;&gt;cities -->\n",
       "<g id=\"edge22\" class=\"edge\">\n",
       "<title>reports&#45;&gt;cities</title>\n",
       "<path fill=\"none\" stroke=\"#000000\" d=\"M2999.5,-181.2909C2999.5,-175.8996 2999.5,-173 2999.5,-173 2999.5,-173 4588.5,-173 4588.5,-173 4588.5,-173 4588.5,-154.5369 4588.5,-154.5369\"/>\n",
       "<polygon fill=\"#000000\" stroke=\"#000000\" points=\"4592.0001,-154.5369 4588.5,-144.5369 4585.0001,-154.537 4592.0001,-154.5369\"/>\n",
       "<text text-anchor=\"middle\" x=\"3783.5861\" y=\"-176.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">City</text>\n",
       "</g>\n",
       "<!-- reports&#45;&gt;zipcodes -->\n",
       "<g id=\"edge23\" class=\"edge\">\n",
       "<title>reports&#45;&gt;zipcodes</title>\n",
       "<path fill=\"none\" stroke=\"#000000\" d=\"M3002.5,-181.3538C3002.5,-178.5021 3002.5,-177 3002.5,-177 3002.5,-177 4759.5,-177 4759.5,-177 4759.5,-177 4759.5,-147.1011 4759.5,-147.1011\"/>\n",
       "<polygon fill=\"#000000\" stroke=\"#000000\" points=\"4763.0001,-147.1011 4759.5,-137.1011 4756.0001,-147.1012 4763.0001,-147.1011\"/>\n",
       "<text text-anchor=\"middle\" x=\"3858.7725\" y=\"-180.8\" font-family=\"Times,serif\" font-size=\"14.00\" fill=\"#000000\">ZiPCODE</text>\n",
       "</g>\n",
       "</g>\n",
       "</svg>\n"
      ],
      "text/plain": [
       "<graphviz.dot.Digraph at 0x7f68f490be80>"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print('entering featuretools...')\n",
    "df['REPORT_ID'] = df['REPORT_ID'].astype(int)\n",
    "\n",
    "es = ft.EntitySet()\n",
    "\n",
    "# entity_id: [entity_index, [entity_variables]]\n",
    "main_mapping = {\n",
    "    'account_and_contributors': ['ACCT_CONTRIBUTOR_ID', {\n",
    "        'ACCT_CONTRIBUTOR_ID': ft.variable_types.Index,\n",
    "    }],\n",
    "    'account_and_ownerships': ['ACCT_OWNERSHIP_ID', {\n",
    "        'ACCT_OWNERSHIP_ID': ft.variable_types.Index,\n",
    "    }],\n",
    "    'contributors_and_ownerships': ['CONTRIBUTOR_OWNERSHIP_ID', {\n",
    "        'CONTRIBUTOR_OWNERSHIP_ID': ft.variable_types.Index,\n",
    "    }],\n",
    "    'agreements': ['ID', {\n",
    "        'ID': ft.variable_types.Index,\n",
    "    }],\n",
    "    'account_types': ['ACCT-TYPE', {\n",
    "        'ACCT-TYPE': ft.variable_types.Index,\n",
    "    }],\n",
    "    'self_indicators': ['SELF-INDICATOR', {\n",
    "        'SELF-INDICATOR': ft.variable_types.Index,\n",
    "    }],\n",
    "    'match_types': ['MATCH-TYPE', {\n",
    "        'MATCH-TYPE': ft.variable_types.Index,\n",
    "    }],\n",
    "    'contributor_types': ['CONTRIBUTOR-TYPE', {\n",
    "        'CONTRIBUTOR-TYPE': ft.variable_types.Index,\n",
    "    }],\n",
    "    'ownerships': ['OWNERSHIP-IND', {\n",
    "        'OWNERSHIP-IND': ft.variable_types.Index,\n",
    "    }],\n",
    "    'account_statuses': ['ACCOUNT-STATUS', {\n",
    "        'ACCOUNT-STATUS': ft.variable_types.Index,\n",
    "    }],\n",
    "    'asset_classes': ['ASSET_CLASS', {\n",
    "        'ASSET_CLASS': ft.variable_types.Index,\n",
    "    }],\n",
    "    'age': ['AGE_BINS', {\n",
    "        'AGE_BINS': ft.variable_types.Index,\n",
    "    }],\n",
    "    'ltv': ['LTV_BINS', {\n",
    "        'LTV_BINS': ft.variable_types.Index,\n",
    "    }],\n",
    "    'frequency': ['Frequency', {\n",
    "        'Frequency': ft.variable_types.Index,\n",
    "    }],\n",
    "    'installment_modes': ['InstlmentMode', {\n",
    "        'InstlmentMode': ft.variable_types.Index,\n",
    "    }],\n",
    "    'loan_status': ['LoanStatus', {\n",
    "        'LoanStatus': ft.variable_types.Index,\n",
    "    }],\n",
    "    'payment_modes': ['PaymentMode', {\n",
    "        'PaymentMode': ft.variable_types.Index,\n",
    "    }],\n",
    "    'branches': ['BranchID', {\n",
    "        'BranchID': ft.variable_types.Index,\n",
    "        'Area': ft.variable_types.Id,\n",
    "    }],\n",
    "    'suppliers': ['SupplierID', {\n",
    "        'SupplierID': ft.variable_types.Index,\n",
    "    }],\n",
    "    'manufacturers': ['ManufacturerID', {\n",
    "        'ManufacturerID': ft.variable_types.Index,\n",
    "    }],\n",
    "    'sex': ['SEX', {\n",
    "        'SEX': ft.variable_types.Index,\n",
    "    }],\n",
    "    'cities': ['City', {\n",
    "        'City': ft.variable_types.Index,\n",
    "        'State': ft.variable_types.Id,\n",
    "    }],\n",
    "    'zipcodes': ['ZiPCODE', {\n",
    "        'ZiPCODE': ft.variable_types.Index, # 444511\n",
    "    }],\n",
    "}\n",
    "\n",
    "branches_mapping = {\n",
    "    'areas': ['Area', {\n",
    "        'Area': ft.variable_types.Index,\n",
    "    }],\n",
    "}\n",
    "\n",
    "cities_mapping = {\n",
    "    'states': ['State', {\n",
    "        'State': ft.variable_types.Index,\n",
    "    }],\n",
    "}\n",
    "\n",
    "main_variables = {\n",
    "    'REPORT_ID' : ft.variable_types.Index,\n",
    "    \n",
    "    'AuthDate': ft.variable_types.Datetime,\n",
    "    'MaturityDAte': ft.variable_types.Datetime,\n",
    "    'DisbursalDate': ft.variable_types.Datetime,\n",
    "    \n",
    "    'CLOSE-DT': ft.variable_types.Datetime,\n",
    "    'DISBURSED-DT': ft.variable_types.Datetime,\n",
    "    'DATE-REPORTED': ft.variable_types.Datetime,\n",
    "    'LAST-PAYMENT-DATE': ft.variable_types.Datetime,\n",
    "}\n",
    "\n",
    "\n",
    "# ==========================================\n",
    "# Creating Level 2 Entities\n",
    "# ==========================================\n",
    "for entity_id, (entity_index, entity_variables) in cities_mapping.items():\n",
    "    df_index = df.drop_duplicates([entity_index])[list(entity_variables)].reset_index(drop=True)\n",
    "    es = es.entity_from_dataframe(entity_id=entity_id, dataframe=df_index, index=entity_index, variable_types=entity_variables)\n",
    "\n",
    "    \n",
    "for entity_id, (entity_index, entity_variables) in branches_mapping.items():\n",
    "    df_index = df.drop_duplicates([entity_index])[list(entity_variables)].reset_index(drop=True)\n",
    "    es = es.entity_from_dataframe(entity_id=entity_id, dataframe=df_index, index=entity_index, variable_types=entity_variables)\n",
    "    \n",
    "# ==========================================\n",
    "# Creating Level 1 Entities\n",
    "# ==========================================\n",
    "for entity_id, (entity_index, entity_variables) in main_mapping.items():\n",
    "    \n",
    "    df_index = df.drop_duplicates([entity_index])[list(entity_variables)].reset_index(drop=True)\n",
    "    es = es.entity_from_dataframe(entity_id=entity_id, dataframe=df_index, index=entity_index, variable_types=entity_variables)\n",
    "    main_variables[entity_index] = ft.variable_types.Id\n",
    "\n",
    "# ==========================================\n",
    "# Adding numeric variables\n",
    "# ==========================================\n",
    "main_variables['Tenure'] = ft.variable_types.Numeric \n",
    "main_variables['AssetCost'] = ft.variable_types.Numeric \n",
    "main_variables['AmountFinance'] = ft.variable_types.Numeric \n",
    "main_variables['DisbursalAmount'] = ft.variable_types.Numeric \n",
    "main_variables['EMI'] = ft.variable_types.Numeric \n",
    "main_variables['LTV'] = ft.variable_types.Numeric \n",
    "main_variables['AGE'] = ft.variable_types.Numeric \n",
    "main_variables['MonthlyIncome'] = ft.variable_types.Numeric \n",
    "\n",
    "for c in additional_con_cols:\n",
    "    main_variables[c] = ft.variable_types.Numeric \n",
    "\n",
    "# ['CLOSE-DT', 'DISBURSED-DT', 'DATE-REPORTED', 'LAST-PAYMENT-DATE'] \n",
    "for c in source_con_cols_seq + additional_con_seq_cols:\n",
    "    main_variables[c] = ft.variable_types.Numeric \n",
    "    \n",
    "\n",
    "for c in additional_cat_seq_cols + additional_cat_cols: \n",
    "    main_variables[c] = ft.variable_types.Categorical\n",
    "\n",
    "main_variables['F_NO_TOP_UP'] = ft.variable_types.Boolean\n",
    "main_variables['F_48_MONTHS'] = ft.variable_types.Boolean\n",
    "main_variables['F_24_30_MONTHS'] = ft.variable_types.Boolean\n",
    "main_variables['F_12_18_MONTHS'] = ft.variable_types.Boolean\n",
    "main_variables['F_18_24_MONTHS'] = ft.variable_types.Boolean\n",
    "main_variables['F_30_36_MONTHS'] = ft.variable_types.Boolean\n",
    "main_variables['F_36_48_MONTHS'] = ft.variable_types.Boolean\n",
    "\n",
    "\n",
    "# ======================\n",
    "# Adding Main Entity\n",
    "# ======================\n",
    "# WARNING: Using only `df_train_dataset` and `df_test_dataset`\n",
    "es = es.entity_from_dataframe(entity_id=\"reports\", dataframe=df[list(main_variables.keys())].copy(), index=\"REPORT_ID\", variable_types=main_variables)\n",
    "\n",
    "# ==========================================\n",
    "# Adding Relationships to Main\n",
    "# ==========================================\n",
    "for entity_id, (entity_index, _) in main_mapping.items():\n",
    "    es = es.add_relationship(ft.Relationship(es[entity_id][entity_index], es[\"reports\"][entity_index]))\n",
    "\n",
    "# ==========================================\n",
    "# Adding Relationships to Level 2\n",
    "# ==========================================\n",
    "for entity_id, (entity_index, _) in cities_mapping.items():\n",
    "    es = es.add_relationship(ft.Relationship(es[entity_id][entity_index], es[\"cities\"][entity_index]))  \n",
    "    \n",
    "for entity_id, (entity_index, _) in branches_mapping.items():\n",
    "    es = es.add_relationship(ft.Relationship(es[entity_id][entity_index], es[\"branches\"][entity_index]))  \n",
    "    \n",
    "es.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "LTFS_LOAN_SPECIFICS_CON = [\n",
    "    'AGE', \n",
    "    'AmountFinance', \n",
    "    'AssetCost', \n",
    "    'DisbursalAmount', \n",
    "    'EMI', \n",
    "    'LTV', \n",
    "    'MonthlyIncome', \n",
    "    'Tenure',\n",
    "] + additional_con_cols\n",
    "\n",
    "LTFS_LOAN_SPECIFICS_CAT = source_cat_cols + additional_cat_cols\n",
    "\n",
    "LTFS_LOAN_SPECIFICS_BOOL = [\n",
    "    'F_12_18_MONTHS',\n",
    "    'F_18_24_MONTHS',\n",
    "    'F_24_30_MONTHS',\n",
    "    'F_30_36_MONTHS',\n",
    "    'F_36_48_MONTHS',\n",
    "    'F_48_MONTHS',\n",
    "    'F_NO_TOP_UP',\n",
    "]\n",
    "\n",
    "BUREAU_LOAN_CAT = [\n",
    "    'ACCT-TYPE', \n",
    "    'SELF-INDICATOR', \n",
    "    'MATCH-TYPE', \n",
    "    'CONTRIBUTOR-TYPE', \n",
    "    'OWNERSHIP-IND', \n",
    "    'ACCOUNT-STATUS', \n",
    "    'ASSET_CLASS',\n",
    "    \n",
    "##  commented as resource expensive\n",
    "#     'ACCT_CONTRIBUTOR_ID',\n",
    "#     'ACCT_OWNERSHIP-ID',\n",
    "#     'CONTRIBUTOR_OWNERSHIP_ID',\n",
    "] + additional_cat_seq_cols"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "id": "AEboGNAyZqgx"
   },
   "outputs": [],
   "source": [
    "from featuretools.primitives.standard.aggregation_primitives import NMostCommon\n",
    "n_most_common = NMostCommon(n=1)\n",
    "\n",
    "ft_args = namedtuple('ft_args', [\n",
    "    'entity_name', \n",
    "    'target_entity', \n",
    "    'target_entity_id',\n",
    "    'ignore_variables',\n",
    "    'ignore_entities',\n",
    "    'agg_primitives',\n",
    "    'trans_primitives',\n",
    "    'interesting_values',\n",
    "    'where_primitives',\n",
    "    'drop_exact',\n",
    "    'max_depth',\n",
    "    'primitive_options',\n",
    "])\n",
    "\n",
    "\n",
    "# ['mean', 'std', 'max', 'min', 'num_unique', 'count', 'percent_true']\n",
    "# https://stackoverflow.com/questions/62193884/use-featuretools-on-specific-columns-only\n",
    "target_entities = [    \n",
    "    ft_args('reports_et', \n",
    "            target_entity='reports', \n",
    "            target_entity_id='REPORT_ID', \n",
    "            ignore_variables=None, \n",
    "            ignore_entities=[], \n",
    "            agg_primitives=[],\n",
    "            trans_primitives=[\n",
    "                'day', \n",
    "                'month', \n",
    "                'year', \n",
    "                'week', \n",
    "                'weekday', \n",
    "                'is_weekend',\n",
    "            ],\n",
    "            interesting_values=[],\n",
    "            where_primitives=[], \n",
    "            drop_exact=list(main_variables.keys()) + [\n",
    "                'branches.Area',\n",
    "                'cities.State',\n",
    "            ],\n",
    "            max_depth=1,\n",
    "            primitive_options=None,\n",
    "            ),\n",
    "     ft_args('agreements_et', \n",
    "            target_entity='agreements', \n",
    "            target_entity_id='ID', \n",
    "            ignore_variables=None, \n",
    "            ignore_entities=[],\n",
    "            agg_primitives=['count', 'num_unique', n_most_common],\n",
    "            trans_primitives=[],\n",
    "            interesting_values=BUREAU_LOAN_CAT,\n",
    "            where_primitives=['count'],\n",
    "            drop_exact=[],\n",
    "            max_depth=2,\n",
    "            primitive_options={\n",
    "                ('num_unique', n_most_common): {\n",
    "                    'ignore_variables': {\n",
    "                        'reports': [c for c in LTFS_LOAN_SPECIFICS_CAT if c not in ['Area', 'State']],\n",
    "                        'branches': ['Area'],\n",
    "                        'cities': ['State'],\n",
    "                    }\n",
    "                }\n",
    "            }\n",
    "    ),\n",
    "    ft_args('account_types_et', \n",
    "            target_entity='account_types', \n",
    "            target_entity_id='ACCT-TYPE', \n",
    "            ignore_variables=None, \n",
    "            ignore_entities=[],\n",
    "            agg_primitives=['count', 'percent_true'],\n",
    "            trans_primitives=[],\n",
    "            interesting_values=[],\n",
    "            where_primitives=[], \n",
    "            drop_exact=[],\n",
    "            max_depth=2,\n",
    "            primitive_options=None\n",
    "    ),\n",
    "    ft_args('contributor_types_et', \n",
    "            target_entity='contributor_types', \n",
    "            target_entity_id='CONTRIBUTOR-TYPE', \n",
    "            ignore_variables=None, \n",
    "            ignore_entities=[],\n",
    "            agg_primitives=['count', 'percent_true'],\n",
    "            trans_primitives=[],\n",
    "            interesting_values=[],\n",
    "            where_primitives=[], \n",
    "            drop_exact=[],\n",
    "            max_depth=2,\n",
    "            primitive_options=None\n",
    "    )\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 270,
     "referenced_widgets": [
      "4fcec6a90d574301b6b65b46505811d1",
      "3f3d59e40eac44d5a73962aba01436f6",
      "2e1557f0fbcc47d7b4fb6703d92ffcbd",
      "e2002845e3984b7bba22197d390c12f6",
      "2560fe543fa84310b7093eaafa085bd7",
      "5ffbf3667e6a4c24acc3a2b39e2184ee",
      "312055e67b444aa79acbf71d9bb22188",
      "f2d69e1ecd3e4062881c5d8ba66ab7ce",
      "485d9098ce37437d9a92301a8b73e0f9",
      "420cbea1c4194e4a948accbb6a9f5f0f",
      "a956dd1d282b405d83e581fe5f2b1997"
     ]
    },
    "id": "D5ditgaiZqgx",
    "outputId": "62cfac50-684c-41e7-feec-627acf1f1172"
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "accc9dbc9c2348a389fa155786cc9fb9",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/4 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Built 42 features\n",
      "Built 126 features\n",
      "Built 8 features\n",
      "Built 8 features\n"
     ]
    }
   ],
   "source": [
    "MAX_DEPTH, N_JOBS = 2, psutil.cpu_count()\n",
    "features = {te.entity_name : dict() for te in target_entities}\n",
    "for te in tqdm(target_entities):\n",
    "  \n",
    "    features[te.entity_name]['id'] = te.target_entity_id\n",
    "    \n",
    "    # assign interesting values only when required by where_primitives\n",
    "    for iv in te.interesting_values:\n",
    "        es['reports'][iv].interesting_values = df[iv].unique().tolist()\n",
    "    \n",
    "    features[te.entity_name]['defs'] = ft.dfs(entityset=es, target_entity=te.target_entity, ignore_entities=te.ignore_entities, \n",
    "                                        ignore_variables=te.ignore_variables, agg_primitives=te.agg_primitives, trans_primitives=te.trans_primitives, \n",
    "                                        max_depth=te.max_depth, drop_exact=te.drop_exact, n_jobs=N_JOBS, verbose=True, primitive_options=te.primitive_options,\n",
    "                                        where_primitives=te.where_primitives, features_only=True)\n",
    "    \n",
    "    # unassign interesting values\n",
    "    for iv in list(es['reports'].variable_types.keys()):\n",
    "        es['reports'][iv].interesting_values = []\n",
    "\n",
    "assert(features['reports_et']['defs'].__len__() == 42)\n",
    "assert(features['agreements_et']['defs'].__len__() == 126)\n",
    "assert(features['account_types_et']['defs'].__len__() == 8)\n",
    "assert(features['contributor_types_et']['defs'].__len__() == 8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 49,
     "referenced_widgets": [
      "137e65f0bfb242e3a694cc22ecf898a1",
      "7777781b712a499f99113a04effcf849",
      "dc4c29e7b3d64f1e9ee6980e36282d75",
      "741b7675afc842c38574c94e42cdaf1f",
      "4d5306188a74469596505b2b8d34c177",
      "64f98a28ab4549609cc166e3052a8deb",
      "f4cb056711424c3081b66cdb1465ed26",
      "61cdefbbf9b04057a166543cf74f4e2c",
      "5abe336638064511bfa086167b2c42e2",
      "66d05b85e9c748969ac9d2dc844d3503",
      "9ba6389e964c4bcd86c8787dcb1357ea"
     ]
    },
    "id": "sP5DmVBMZqgy",
    "outputId": "1563b584-964f-4ea2-921c-feda67a4d141",
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "41cce7847f5f4449b39ea1519dfaff7e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/4 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "features = {te.entity_name : dict() for te in target_entities}\n",
    "for te in tqdm(target_entities):\n",
    "  \n",
    "    file_path = Path(f'{path}/ftools_{te.entity_name}.pkl')\n",
    "    if os.path.isfile(file_path):\n",
    "        continue\n",
    "        \n",
    "    features[te.entity_name]['id'] = te.target_entity_id\n",
    "    \n",
    "    # assign interesting values only when required by where_primitives\n",
    "    for iv in te.interesting_values:\n",
    "        es['reports'][iv].interesting_values = df[iv].unique().tolist()\n",
    "        \n",
    "    features[te.entity_name]['features'], features[te.entity_name]['defs'] = ft.dfs(entityset=es, target_entity=te.target_entity, ignore_entities=te.ignore_entities, \n",
    "                                                            ignore_variables=te.ignore_variables, agg_primitives=te.agg_primitives, trans_primitives=te.trans_primitives, \n",
    "                                                            max_depth=te.max_depth, drop_exact=te.drop_exact, n_jobs=N_JOBS, verbose=True, primitive_options=te.primitive_options,\n",
    "                                                            where_primitives=te.where_primitives, features_only=False)\n",
    "                                                            \n",
    "    # unassign interesting values\n",
    "    for iv in list(es['reports'].variable_types.keys()):\n",
    "        es['reports'][iv].interesting_values = []\n",
    "\n",
    "    column_names = None\n",
    "    if te.entity_name == 'reports_et':\n",
    "        column_names = [f\"{te.entity_name}_con_{c}\" for c in features[te.entity_name]['features'].columns]\n",
    "    elif te.entity_name == 'agreements_et':\n",
    "        suffix =['cat' if 'N_MOST_COMMON' in c else 'con' for c in features[te.entity_name]['features'].columns]\n",
    "        column_names = [f\"{te.entity_name}_{suffix[i]}_{c}\" for i, c in enumerate(features[te.entity_name]['features'].columns)]\n",
    "    elif te.entity_name in ['account_types_et', 'contributor_types_et']:\n",
    "        column_names = [f\"{te.entity_name}_con_{c}\" for c in features[te.entity_name]['features'].columns]\n",
    "                        \n",
    "    assert(column_names is not None), \"column_names are None\"\n",
    "    \n",
    "    # replaceing json sensative chars\n",
    "    column_names  = [re.sub(',', '', c) for c in column_names]\n",
    "    \n",
    "    features[te.entity_name]['features'].columns = column_names\n",
    "    features[te.entity_name]['features'].to_pickle(file_path)\n",
    "    ft.save_features(features[te.entity_name]['defs'], Path(f'{path}/{te.entity_name}_defs.json').__str__())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "id": "8fgkC5B5Zqgy"
   },
   "outputs": [],
   "source": [
    "features = {te.entity_name : dict() for te in target_entities}\n",
    "for te in target_entities:\n",
    "    features[te.entity_name]['id'] = te.target_entity_id\n",
    "    features[te.entity_name]['features'] = pd.read_pickle(f'{path}/ftools_{te.entity_name}.pkl')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 270,
     "referenced_widgets": [
      "839ea94b4da3447090066fe7a1ea44bc",
      "073eeeb555194babb0912cd0ea61a486",
      "41caf7e06b7540e7a7e2cd29911ac69a",
      "cc4e2d90f21d4a3abce359e04c86915c",
      "2ee3bf088f1743798c2cb34f5b7d7c33",
      "4e91aabcab994820a94412a1506385b6",
      "9ef3f263da144e3da8a68f20f7c0cec1",
      "286fd37aa66d4d66b8ce5170ef0d49e4",
      "8072c2d6360342f2a5e21ea7bf730845",
      "d6af4609c7e0470ba85701fb23f7c59a",
      "d025956149504e0cb82efa976b76548b"
     ]
    },
    "id": "9Mn_iNX1Zqgy",
    "outputId": "bee8cbe7-a34e-450f-fa1f-aeb894cbee8e"
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "8d30ac3d23ed4f8eabd2676c0e117977",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/4 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "merging... reports_et\n",
      "merging... agreements_et\n",
      "merging... account_types_et\n",
      "merging... contributor_types_et\n"
     ]
    }
   ],
   "source": [
    "for te_name, te_features in tqdm(features.items()):\n",
    "    print('merging...', te_name)\n",
    "    df = df.merge(features[te_name]['features'], left_on=features[te_name]['id'], right_on=features[te_name]['id'], how='left', suffixes=('_suffix', ''))\n",
    "    df = df.drop([c for c in df.columns if '_suffix' in c], axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "ft_cat_cols = [c for c in df.columns if 'N_MOST_COMMON' in c and c not in source_cat_cols + source_cat_cols_seq + additional_cat_cols + additional_cat_seq_cols]\n",
    "ft_con_cols = [d for d in [c for c in df.columns if '_con_' in c] if d not in ft_cat_cols + source_con_cols + source_con_cols_seq + additional_con_cols + additional_con_seq_cols]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "segregating columns...\n"
     ]
    }
   ],
   "source": [
    "print('segregating columns...')\n",
    "\n",
    "con_cols_seq = source_con_cols_seq + additional_con_seq_cols + source_dt_cols_seq\n",
    "cat_cols_seq = source_cat_cols_seq + additional_cat_seq_cols\n",
    "\n",
    "con_cols = source_con_cols + additional_con_cols + ft_con_cols + source_dt_cols\n",
    "cat_cols = source_cat_cols + additional_cat_cols + ft_cat_cols"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "H71DKyTxxdr9",
    "outputId": "4eac7858-5ffe-4b11-803e-ba60bc5c5bfe"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "encoding target...\n"
     ]
    }
   ],
   "source": [
    "print(\"encoding target...\")\n",
    "target_encoder = LabelEncoder()\n",
    "df.loc[df['source'] == 'train', 'Top-up Month'] = target_encoder.fit_transform(df[df['source'] == 'train']['Top-up Month'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "A4fRHJe81v_z",
    "outputId": "8e1c78d5-756d-4cad-cd12-70103aa1fbaa"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "encoding date and categorical features...\n"
     ]
    }
   ],
   "source": [
    "print(\"encoding date and categorical features...\")\n",
    "df[cat_cols + cat_cols_seq] = df[cat_cols + cat_cols_seq].fillna(DEFAULT_CATEGORICAL_FILL)\n",
    "\n",
    "fucnT = FunctionTransformer(lambda X: X.astype('U'))\n",
    "df[cat_cols + cat_cols_seq] = fucnT.fit_transform(df[cat_cols + cat_cols_seq])\n",
    "\n",
    "oe = OrdinalEncoder(return_df=False, handle_missing='return_nan', handle_unknown='error', drop_invariant=False)\n",
    "df[cat_cols + cat_cols_seq] = oe.fit_transform(df[cat_cols + cat_cols_seq]) - 1\n",
    "\n",
    "fucnT = FunctionTransformer(lambda X: X.astype('int'))\n",
    "df[cat_cols + cat_cols_seq] = fucnT.fit_transform(df[cat_cols + cat_cols_seq])\n",
    "\n",
    "for col in source_dt_cols + source_dt_cols_seq:\n",
    "    df[col] = df[col].map(date_to_integer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "id": "5cIiDZ303hSe"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "replacing infinity (if any) with np.nan...\n"
     ]
    }
   ],
   "source": [
    "print('replacing infinity (if any) with np.nan...')\n",
    "df[con_cols + con_cols_seq] = df[con_cols + con_cols_seq].replace([np.inf, -np.inf], np.nan)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "filling continouous...\n"
     ]
    }
   ],
   "source": [
    "print('filling continouous...')\n",
    "imputer = MissingIndicator(missing_values=np.nan, features='missing-only', error_on_new=False)\n",
    "df_missing = pd.DataFrame(imputer.fit_transform(df[con_cols].values).astype('int'))\n",
    "\n",
    "missing_columns = [f'missing_flag_con_{i}' for i in range(df_missing.columns.__len__())]\n",
    "df_missing.columns = missing_columns\n",
    "df = df.merge(df_missing, left_index=True, right_index=True, how='inner', suffixes=('_suffix', ''))\n",
    "df = df.drop([c for c in df.columns if '_suffix' in c], axis=1)\n",
    "\n",
    "con_cols = source_con_cols + additional_con_cols + ft_con_cols + source_dt_cols + missing_columns\n",
    "\n",
    "imputer = SimpleImputer(missing_values=np.nan, strategy='median', add_indicator=False)\n",
    "df[con_cols] = imputer.fit_transform(df[con_cols].values)\n",
    "\n",
    "df[con_cols_seq] = df[con_cols_seq].fillna(0.)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "rescaling continuous...\n"
     ]
    }
   ],
   "source": [
    "print('rescaling continuous...')\n",
    "scaler = RobustScaler()\n",
    "df[con_cols + con_cols_seq] = scaler.fit_transform(df[con_cols + con_cols_seq].values)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['cat_cols_seq.joblib']"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# df.to_pickle('checkpoint.pkl')\n",
    "# joblib.dump(con_cols, 'con_cols.joblib')\n",
    "# joblib.dump(cat_cols, 'cat_cols.joblib')\n",
    "# joblib.dump(target_encoder, 'target_encoder.joblib')\n",
    "# joblib.dump(con_cols_seq, 'con_cols_seq.joblib')\n",
    "# joblib.dump(cat_cols_seq, 'cat_cols_seq.joblib')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_pickle('checkpoint.pkl')\n",
    "con_cols_seq, cat_cols_seq = joblib.load('con_cols_seq.joblib'), joblib.load('cat_cols_seq.joblib')\n",
    "con_cols, cat_cols, target_encoder = joblib.load('con_cols.joblib'), joblib.load('cat_cols.joblib'), joblib.load('target_encoder.joblib')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "sequence_dim = dict()\n",
    "\n",
    "def seq_slot0(df, sequence_columns):\n",
    "    \"\"\"\n",
    "    https://stackoverflow.com/questions/39674713/neural-network-lstm-input-shape-from-dataframe\n",
    "    \"\"\"\n",
    "    df = df.sort_values('DISBURSED-DT')\n",
    "    df = df.drop_duplicates('DISBURSED-DT')\n",
    "    df['seq'] = df[sequence_columns].apply(tuple, axis=1).apply(list).apply(lambda x: [list(x)]).cumsum()\n",
    "    return df\n",
    "\n",
    "def build_sequence_data(df, essential_columns, group_col, final_col, max_sequence_length=None):\n",
    "\n",
    "    sequence_columns = [c for c in essential_columns if c not in ['DISBURSED-DT', group_col]]\n",
    "    df_grp = df[essential_columns].groupby([group_col]).progress_apply(seq_slot0, sequence_columns=sequence_columns).reset_index(drop=True)\n",
    "\n",
    "    if max_sequence_length is None:\n",
    "        max_sequence_length = df_grp['seq'].apply(len).max()\n",
    "\n",
    "    dimensions = sequence_columns.__len__()\n",
    "    return df_grp, max_sequence_length, dimensions\n",
    "\n",
    "\n",
    "essential_columns = ['ID'] + con_cols_seq\n",
    "group_col= 'ID'\n",
    "final_col = 'p_con_seq_0'\n",
    "\n",
    "file_path = f'{path}/df_{final_col}.pkl'\n",
    "if os.path.isfile(file_path):\n",
    "    df_grp = pd.read_pickle(file_path)\n",
    "    max_sequence_length = joblib.load(f'{path}/max_sequence_length.joblib')\n",
    "    dimensions = joblib.load(f'{path}/dimensions.joblib')\n",
    "else:\n",
    "    df_grp, max_sequence_length, dimensions = build_sequence_data(df, essential_columns, group_col, final_col)\n",
    "    df_grp.to_pickle(file_path)\n",
    "    joblib.dump(max_sequence_length, f'{path}/max_sequence_length.joblib')\n",
    "    joblib.dump(dimensions, f'{path}/dimensions.joblib')\n",
    "    \n",
    "df_grp[final_col] = df_grp['seq'].map(lambda s: np.squeeze(pad_sequences([s], max_sequence_length, dtype='float32'))).apply(np.asarray)\n",
    "df = df.merge(df_grp.drop([c for c in essential_columns if c not in ['DISBURSED-DT', group_col]] + ['seq'], axis=1), on=['DISBURSED-DT', group_col], how='inner', suffixes=('_suffix', ''))\n",
    "df = df.drop([c for c in df.columns if '_suffix' in c], axis=1)\n",
    "del df_grp\n",
    "\n",
    "sequence_dim[final_col] = (max_sequence_length, dimensions)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "modelling...\n"
     ]
    }
   ],
   "source": [
    "print('modelling...')\n",
    "train, test = df[df['source'] == 'train'], df[df['source'] == 'test']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "def run_KERAS(params, train, test, feature_names, con_cols, cat_cols, categorical_dim, sequence_dim, n_folds=10, seed=0, return_models=False):\n",
    "    \"\"\"\n",
    "    https://stackoverflow.com/questions/56303572/how-can-i-solve-this-unknown-label-type-error\n",
    "    \"\"\"\n",
    "    skf = GroupKFold(n_splits=n_folds)\n",
    "    X, y, groups = train[feature_names], train['Top-up Month'].values.astype('int'), train['ID'].values.astype('int')\n",
    "\n",
    "    lstm_cols = list(sequence_dim.keys())\n",
    "    num_classes = np.unique(y).size\n",
    "    preds = np.zeros((test.shape[0], num_classes))\n",
    "\n",
    "    X_test_dl = get_categorical_inputs_for_dl(test[cat_cols].values)\n",
    "    X_test_dl.append(test[con_cols].values)\n",
    "#     X_test_dl = list()\n",
    "    for lc in lstm_cols:\n",
    "        X_test_dl.append([np.hstack(np.asarray(test[lc] )).reshape(len(test[lc]), sequence_dim[lc][0], sequence_dim[lc][1])])\n",
    "    \n",
    "    models = list()\n",
    "    for i, (train_index, test_index) in enumerate(skf.split(X, y, groups)):\n",
    "        X_train, X_val = X.iloc[train_index, :], X.iloc[test_index, :]\n",
    "        y_train, y_val = y[train_index], y[test_index]\n",
    "\n",
    "        y_train = to_categorical(y_train, num_classes=num_classes)\n",
    "        y_val = to_categorical(y_val, num_classes=num_classes)\n",
    "        \n",
    "        # preparing data\n",
    "        X_train_dl = get_categorical_inputs_for_dl(X_train[cat_cols].values)\n",
    "        X_val_dl = get_categorical_inputs_for_dl(X_val[cat_cols].values)\n",
    "        X_train_dl.append(X_train[con_cols].values)\n",
    "        X_val_dl.append(X_val[con_cols].values)\n",
    "        \n",
    "#         X_train_dl, X_val_dl = [], []\n",
    "        for lc in lstm_cols:\n",
    "            X_train_dl.append([np.hstack(np.asarray(X_train[lc])).reshape(len(X_train[lc]), sequence_dim[lc][0], sequence_dim[lc][1])])\n",
    "            X_val_dl.append([np.hstack(np.asarray(X_val[lc])).reshape(len(X_val[lc]), sequence_dim[lc][0], sequence_dim[lc][1])])\n",
    "\n",
    "        terminate_onnan = TerminateOnNaN()\n",
    "        early_stop = EarlyStopping(monitor=\"val_loss\", patience=2, mode='min') \n",
    "        check_pointer = ModelCheckpoint(save_best_only=True, filepath='models/model.{epoch:02d}-{val_loss:.6f}.h5', mode='min', monitor=\"val_loss\", verbose=1)\n",
    "\n",
    "        print('building model...')\n",
    "        \n",
    "        model = build_model(con_cols, categorical_dim, sequence_dim)\n",
    "        history = model.fit(\n",
    "            X_train_dl, \n",
    "            y_train, \n",
    "            validation_data=(X_val_dl, y_val), \n",
    "            epochs=2, \n",
    "            callbacks=[terminate_onnan, early_stop, check_pointer], \n",
    "            verbose=1, **params)\n",
    "        \n",
    "        # loading best\n",
    "        list_of_files = glob.glob('models/*.h5')\n",
    "        latest_file = max(list_of_files, key=os.path.getctime)\n",
    "        print(f'Using model file: [{latest_file}]')    \n",
    "        bst = load_model(latest_file)\n",
    "\n",
    "        test_preds = bst.predict(X_test_dl)\n",
    "        preds += test_preds\n",
    "        models.append(bst)\n",
    "        \n",
    "    if return_models:\n",
    "        return preds / n_folds, models\n",
    "    return preds / n_folds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# del X\n",
    "import gc\n",
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "def build_model(continuous, categoricals_dim, sequence_dim):\n",
    "    \n",
    "    K.clear_session()\n",
    "    continuous_inputs = [Input(shape=(continuous.__len__(),))]\n",
    "    embedding_inputs, embedding_outputs = get_embeddings(categoricals_dim)\n",
    "    \n",
    "    seq_inputs=[Input(shape=(sequence_dim['p_con_seq_0'][0], sequence_dim['p_con_seq_0'][1],))]\n",
    "    \n",
    "    # ========================\n",
    "    # LSTM MODULE\n",
    "    # ========================   \n",
    "    model_seq = seq_inputs[0]\n",
    "    model_seq = LSTM(32, dropout=0.2, recurrent_dropout=0.2)(model_seq)\n",
    "#     model_seq = Convolution1D(16, kernel_size=8, padding='causal', dilation_rate=1)(model_seq)\n",
    "#     model_seq = Convolution1D(16, kernel_size=8, padding='causal', dilation_rate=2)(model_seq)\n",
    "#     model_seq = Convolution1D(16, kernel_size=8, padding='causal', dilation_rate=4)(model_seq)\n",
    "#     model_seq = Convolution1D(16, kernel_size=8, padding='causal', dilation_rate=8)(model_seq)\n",
    "#     model_seq = GlobalMaxPooling1D()(model_seq)\n",
    "    \n",
    "    # ========================\n",
    "    # FCNN MODULE\n",
    "    # ========================   \n",
    "    model_unit = Concatenate()(embedding_outputs + continuous_inputs + [model_seq])\n",
    "    model_unit = Dense(1024, activation='relu', kernel_initializer='he_normal')(model_unit) # kernel_regularizer=l2(1e-2), \n",
    "#     model_unit = BatchNormalization()(model_unit)\n",
    "    model_unit = Dropout(0.4)(model_unit)\n",
    "#     model_unit = Dense(256, activation='relu', kernel_initializer='he_normal')(model_unit) # kernel_regularizer=l2(1e-2), \n",
    "#     model_unit = Dropout(0.3)(model_unit)\n",
    "    model_unit = Dense(256, activation='relu', kernel_initializer='he_normal')(model_unit) # kernel_regularizer=l2(1e-2), \n",
    "    model_outputs = Dense(7, activation='softmax')(model_unit)\n",
    "    \n",
    "    optim = Adam(learning_rate=0.001)\n",
    "    model = Model(inputs=embedding_inputs + continuous_inputs + seq_inputs, outputs=model_outputs)\n",
    "    model.compile(loss='categorical_crossentropy', optimizer=optim, metrics=['accuracy'])\n",
    "    \n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "WRG0-k1fxdsF",
    "outputId": "be18ecb6-655b-4624-b2a5-5f247d4b9a2d",
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "building model...\n",
      "Epoch 1/2\n",
      "877/877 [==============================] - 148s 164ms/step - loss: 68.9335 - accuracy: 0.4910 - val_loss: 8.2538 - val_accuracy: 0.5647\n",
      "\n",
      "Epoch 00001: val_loss improved from inf to 8.25377, saving model to models/model.01-8.253774.h5\n",
      "Epoch 2/2\n",
      "877/877 [==============================] - 143s 163ms/step - loss: 10.8081 - accuracy: 0.7352 - val_loss: 1.8861 - val_accuracy: 0.5669\n",
      "\n",
      "Epoch 00002: val_loss improved from 8.25377 to 1.88612, saving model to models/model.02-1.886122.h5\n",
      "Using model file: [models/model.02-1.886122.h5]\n",
      "building model...\n",
      "Epoch 1/2\n",
      "877/877 [==============================] - 149s 165ms/step - loss: 54.3861 - accuracy: 0.4990 - val_loss: 7.2970 - val_accuracy: 0.5170\n",
      "\n",
      "Epoch 00001: val_loss improved from inf to 7.29698, saving model to models/model.01-7.296984.h5\n",
      "Epoch 2/2\n",
      "877/877 [==============================] - 143s 163ms/step - loss: 2.3797 - accuracy: 0.7360 - val_loss: 2.1130 - val_accuracy: 0.5591\n",
      "\n",
      "Epoch 00002: val_loss improved from 7.29698 to 2.11296, saving model to models/model.02-2.112958.h5\n",
      "Using model file: [models/model.02-2.112958.h5]\n",
      "building model...\n",
      "Epoch 1/2\n",
      "877/877 [==============================] - 150s 165ms/step - loss: 77.3582 - accuracy: 0.4983 - val_loss: 12.7886 - val_accuracy: 0.5105\n",
      "\n",
      "Epoch 00001: val_loss improved from inf to 12.78856, saving model to models/model.01-12.788560.h5\n",
      "Epoch 2/2\n",
      "877/877 [==============================] - 143s 163ms/step - loss: 8.8729 - accuracy: 0.7330 - val_loss: 1.9694 - val_accuracy: 0.5645\n",
      "\n",
      "Epoch 00002: val_loss improved from 12.78856 to 1.96936, saving model to models/model.02-1.969359.h5\n",
      "Using model file: [models/model.02-1.969359.h5]\n",
      "building model...\n",
      "Epoch 1/2\n",
      "877/877 [==============================] - 149s 164ms/step - loss: 92.1039 - accuracy: 0.4944 - val_loss: 42.2991 - val_accuracy: 0.5287\n",
      "\n",
      "Epoch 00001: val_loss improved from inf to 42.29913, saving model to models/model.01-42.299129.h5\n",
      "Epoch 2/2\n",
      "877/877 [==============================] - 144s 164ms/step - loss: 26.8969 - accuracy: 0.7308 - val_loss: 1.9294 - val_accuracy: 0.5659\n",
      "\n",
      "Epoch 00002: val_loss improved from 42.29913 to 1.92940, saving model to models/model.02-1.929402.h5\n",
      "Using model file: [models/model.02-1.929402.h5]\n",
      "building model...\n",
      "Epoch 1/2\n",
      "877/877 [==============================] - 151s 166ms/step - loss: 76.9396 - accuracy: 0.4764 - val_loss: 8.7690 - val_accuracy: 0.5358\n",
      "\n",
      "Epoch 00001: val_loss improved from inf to 8.76904, saving model to models/model.01-8.769042.h5\n",
      "Epoch 2/2\n",
      "877/877 [==============================] - 144s 165ms/step - loss: 10.5059 - accuracy: 0.7230 - val_loss: 1.9332 - val_accuracy: 0.5710\n",
      "\n",
      "Epoch 00002: val_loss improved from 8.76904 to 1.93325, saving model to models/model.02-1.933249.h5\n",
      "Using model file: [models/model.02-1.933249.h5]\n"
     ]
    }
   ],
   "source": [
    "params = {\n",
    "    \"batch_size\": 512,\n",
    "    \"shuffle\": True, \n",
    "    \"workers\" : psutil.cpu_count(),\n",
    "    \"class_weight\": {k: v for k, v in zip(range(7), compute_class_weight('balanced', classes=list(range(7)), y=train['Top-up Month'].values.astype('int')))}\n",
    "}\n",
    "\n",
    "lstm_cols = list(sequence_dim.keys())\n",
    "categorical_dim = [(c, df[c].unique().size) for c in cat_cols]\n",
    "predictors = con_cols + cat_cols + lstm_cols\n",
    "preds_, models = run_KERAS(\n",
    "    params, \n",
    "    train, \n",
    "    test, \n",
    "    predictors, \n",
    "    con_cols=con_cols, \n",
    "    cat_cols=cat_cols, \n",
    "    categorical_dim=categorical_dim, \n",
    "    sequence_dim=sequence_dim, \n",
    "    n_folds=5, \n",
    "    seed=21, \n",
    "    return_models=True\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "id": "55txGvImxdsJ"
   },
   "outputs": [],
   "source": [
    "test['final_preds'] = target_encoder.inverse_transform(np.argmax(preds_, axis=1))\n",
    "\n",
    "df_submit = test.groupby('ID')['final_preds'].agg(lambda x:x.value_counts().index[0]).to_frame().reset_index()\n",
    "df_submit.columns = ['ID', 'Top-up Month']\n",
    "df_submit.to_csv(f\"submissions/submission_{datetime.now().strftime('%Y%m%d%H%M%S')}.csv\", index = False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "XWyPCIBZxdsL"
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "solution.ipynb",
   "provenance": [],
   "toc_visible": true
  },
  "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": "3.6.8"
  },
  "widgets": {
   "application/vnd.jupyter.widget-state+json": {
    "01b94cc94e7c44a2b6f53cb461aebed2": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_8b7a415065d3443aa8704cc739e46dd7",
      "placeholder": "​",
      "style": "IPY_MODEL_5643a581c5194bd091427206c1a99dda",
      "value": "100%"
     }
    },
    "0414be820ad444598fd5256b08e6fb82": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "073eeeb555194babb0912cd0ea61a486": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "126bf385d40142079e9f95418a3f3bcc": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_a45977e44c244609b764d64773715c72",
      "max": 1,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_a2cdfc3571a640e8b977706816b5b889",
      "value": 1
     }
    },
    "137e65f0bfb242e3a694cc22ecf898a1": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_dc4c29e7b3d64f1e9ee6980e36282d75",
       "IPY_MODEL_741b7675afc842c38574c94e42cdaf1f",
       "IPY_MODEL_4d5306188a74469596505b2b8d34c177"
      ],
      "layout": "IPY_MODEL_7777781b712a499f99113a04effcf849"
     }
    },
    "1bb3243cb8be4494979b511ed0289b53": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_01b94cc94e7c44a2b6f53cb461aebed2",
       "IPY_MODEL_1cd535b3d6244892bcc1e4cd55f2a28a",
       "IPY_MODEL_d056cc7a5a1c45f2b2da5ac910833a82"
      ],
      "layout": "IPY_MODEL_85becb2b2fca4edd85137ae368a5cd0f"
     }
    },
    "1cd535b3d6244892bcc1e4cd55f2a28a": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_999edd228bd74e4a84d4e6e622a0baaf",
      "max": 1,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_d98023b987584879b387ab579d533dc7",
      "value": 1
     }
    },
    "1cd80aa6e7a64856aebb169135f54e99": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_a0a60f2219b149f9b1165fe0b19f3585",
      "placeholder": "​",
      "style": "IPY_MODEL_e357505f76ef45d29dc2fda93eb905d9",
      "value": " 1/1 [00:00&lt;00:00,  3.40it/s]"
     }
    },
    "2560fe543fa84310b7093eaafa085bd7": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_a956dd1d282b405d83e581fe5f2b1997",
      "placeholder": "​",
      "style": "IPY_MODEL_420cbea1c4194e4a948accbb6a9f5f0f",
      "value": " 13/13 [00:00&lt;00:00, 23.72it/s]"
     }
    },
    "286fd37aa66d4d66b8ce5170ef0d49e4": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": ""
     }
    },
    "2e1557f0fbcc47d7b4fb6703d92ffcbd": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_312055e67b444aa79acbf71d9bb22188",
      "placeholder": "​",
      "style": "IPY_MODEL_5ffbf3667e6a4c24acc3a2b39e2184ee",
      "value": "100%"
     }
    },
    "2ee3bf088f1743798c2cb34f5b7d7c33": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_d025956149504e0cb82efa976b76548b",
      "placeholder": "​",
      "style": "IPY_MODEL_d6af4609c7e0470ba85701fb23f7c59a",
      "value": " 13/13 [00:18&lt;00:00,  2.72s/it]"
     }
    },
    "312055e67b444aa79acbf71d9bb22188": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "342934a8b36d43c58b63b574d671c90a": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_b969767ff58a41b38734ae8e738ed893",
      "placeholder": "​",
      "style": "IPY_MODEL_8b70f5167f064faaafc1e6567b0b5772",
      "value": "100%"
     }
    },
    "3f3d59e40eac44d5a73962aba01436f6": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "41caf7e06b7540e7a7e2cd29911ac69a": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_9ef3f263da144e3da8a68f20f7c0cec1",
      "placeholder": "​",
      "style": "IPY_MODEL_4e91aabcab994820a94412a1506385b6",
      "value": "100%"
     }
    },
    "420cbea1c4194e4a948accbb6a9f5f0f": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "449bb81eda5b462193d06389bfb160bb": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "465caa8a4ff04ded84d47df52dd0d9c3": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "485d9098ce37437d9a92301a8b73e0f9": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "4d5306188a74469596505b2b8d34c177": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_9ba6389e964c4bcd86c8787dcb1357ea",
      "placeholder": "​",
      "style": "IPY_MODEL_66d05b85e9c748969ac9d2dc844d3503",
      "value": " 13/13 [00:00&lt;00:00, 490.78it/s]"
     }
    },
    "4e91aabcab994820a94412a1506385b6": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "4fcec6a90d574301b6b65b46505811d1": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_2e1557f0fbcc47d7b4fb6703d92ffcbd",
       "IPY_MODEL_e2002845e3984b7bba22197d390c12f6",
       "IPY_MODEL_2560fe543fa84310b7093eaafa085bd7"
      ],
      "layout": "IPY_MODEL_3f3d59e40eac44d5a73962aba01436f6"
     }
    },
    "5643a581c5194bd091427206c1a99dda": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "5abe336638064511bfa086167b2c42e2": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "5ffbf3667e6a4c24acc3a2b39e2184ee": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "61cdefbbf9b04057a166543cf74f4e2c": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": ""
     }
    },
    "64f98a28ab4549609cc166e3052a8deb": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "66d05b85e9c748969ac9d2dc844d3503": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "6fe125d347524667adf33ce6c2dc3401": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "700594d27b98446a8543059cf7288b31": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_c1f6cb9e24714bdca95cfa72f4e96859",
       "IPY_MODEL_126bf385d40142079e9f95418a3f3bcc",
       "IPY_MODEL_1cd80aa6e7a64856aebb169135f54e99"
      ],
      "layout": "IPY_MODEL_0414be820ad444598fd5256b08e6fb82"
     }
    },
    "741b7675afc842c38574c94e42cdaf1f": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_5abe336638064511bfa086167b2c42e2",
      "max": 13,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_61cdefbbf9b04057a166543cf74f4e2c",
      "value": 13
     }
    },
    "7777781b712a499f99113a04effcf849": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "8072c2d6360342f2a5e21ea7bf730845": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "839ea94b4da3447090066fe7a1ea44bc": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_41caf7e06b7540e7a7e2cd29911ac69a",
       "IPY_MODEL_cc4e2d90f21d4a3abce359e04c86915c",
       "IPY_MODEL_2ee3bf088f1743798c2cb34f5b7d7c33"
      ],
      "layout": "IPY_MODEL_073eeeb555194babb0912cd0ea61a486"
     }
    },
    "85becb2b2fca4edd85137ae368a5cd0f": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "8b70f5167f064faaafc1e6567b0b5772": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "8b7a415065d3443aa8704cc739e46dd7": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "8f0ebc9022e942bfa4f73e7799891d65": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_a807ee1b76bb4386adedf1b7f3f11776",
      "placeholder": "​",
      "style": "IPY_MODEL_e7feb8caf5604f8fa2e0f8290a88eb68",
      "value": " 1/1 [00:05&lt;00:00,  5.43s/it]"
     }
    },
    "9082ac3c04a24c0ca764bb3c3e6989e8": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "9391e08a692e44cdbf3222208fe373cb": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": ""
     }
    },
    "999edd228bd74e4a84d4e6e622a0baaf": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "9ba6389e964c4bcd86c8787dcb1357ea": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "9ef3f263da144e3da8a68f20f7c0cec1": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "a0a60f2219b149f9b1165fe0b19f3585": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "a2cdfc3571a640e8b977706816b5b889": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": ""
     }
    },
    "a45977e44c244609b764d64773715c72": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "a807ee1b76bb4386adedf1b7f3f11776": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "a956dd1d282b405d83e581fe5f2b1997": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "b969767ff58a41b38734ae8e738ed893": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "be7f1c9303f44b00bccfed619c1d2215": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "c1f6cb9e24714bdca95cfa72f4e96859": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_d4789c88388c445285a5fa1061fa3559",
      "placeholder": "​",
      "style": "IPY_MODEL_465caa8a4ff04ded84d47df52dd0d9c3",
      "value": "100%"
     }
    },
    "cbb82ed3f16d438fbfcefe760c4a6593": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HBoxModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HBoxModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HBoxView",
      "box_style": "",
      "children": [
       "IPY_MODEL_342934a8b36d43c58b63b574d671c90a",
       "IPY_MODEL_f8bdf383b3f24d208c0d0abf9b7f7f37",
       "IPY_MODEL_8f0ebc9022e942bfa4f73e7799891d65"
      ],
      "layout": "IPY_MODEL_449bb81eda5b462193d06389bfb160bb"
     }
    },
    "cc4e2d90f21d4a3abce359e04c86915c": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_8072c2d6360342f2a5e21ea7bf730845",
      "max": 13,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_286fd37aa66d4d66b8ce5170ef0d49e4",
      "value": 13
     }
    },
    "d025956149504e0cb82efa976b76548b": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "d056cc7a5a1c45f2b2da5ac910833a82": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_6fe125d347524667adf33ce6c2dc3401",
      "placeholder": "​",
      "style": "IPY_MODEL_9082ac3c04a24c0ca764bb3c3e6989e8",
      "value": " 1/1 [00:00&lt;00:00, 33.70it/s]"
     }
    },
    "d4789c88388c445285a5fa1061fa3559": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "d6af4609c7e0470ba85701fb23f7c59a": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "d98023b987584879b387ab579d533dc7": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": ""
     }
    },
    "dc4c29e7b3d64f1e9ee6980e36282d75": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "HTMLModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "HTMLModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "HTMLView",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_f4cb056711424c3081b66cdb1465ed26",
      "placeholder": "​",
      "style": "IPY_MODEL_64f98a28ab4549609cc166e3052a8deb",
      "value": "100%"
     }
    },
    "e2002845e3984b7bba22197d390c12f6": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_485d9098ce37437d9a92301a8b73e0f9",
      "max": 13,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_f2d69e1ecd3e4062881c5d8ba66ab7ce",
      "value": 13
     }
    },
    "e357505f76ef45d29dc2fda93eb905d9": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "e7feb8caf5604f8fa2e0f8290a88eb68": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "DescriptionStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "DescriptionStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "description_width": ""
     }
    },
    "f2d69e1ecd3e4062881c5d8ba66ab7ce": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "ProgressStyleModel",
     "state": {
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "ProgressStyleModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "StyleView",
      "bar_color": null,
      "description_width": ""
     }
    },
    "f4cb056711424c3081b66cdb1465ed26": {
     "model_module": "@jupyter-widgets/base",
     "model_name": "LayoutModel",
     "state": {
      "_model_module": "@jupyter-widgets/base",
      "_model_module_version": "1.2.0",
      "_model_name": "LayoutModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/base",
      "_view_module_version": "1.2.0",
      "_view_name": "LayoutView",
      "align_content": null,
      "align_items": null,
      "align_self": null,
      "border": null,
      "bottom": null,
      "display": null,
      "flex": null,
      "flex_flow": null,
      "grid_area": null,
      "grid_auto_columns": null,
      "grid_auto_flow": null,
      "grid_auto_rows": null,
      "grid_column": null,
      "grid_gap": null,
      "grid_row": null,
      "grid_template_areas": null,
      "grid_template_columns": null,
      "grid_template_rows": null,
      "height": null,
      "justify_content": null,
      "justify_items": null,
      "left": null,
      "margin": null,
      "max_height": null,
      "max_width": null,
      "min_height": null,
      "min_width": null,
      "object_fit": null,
      "object_position": null,
      "order": null,
      "overflow": null,
      "overflow_x": null,
      "overflow_y": null,
      "padding": null,
      "right": null,
      "top": null,
      "visibility": null,
      "width": null
     }
    },
    "f8bdf383b3f24d208c0d0abf9b7f7f37": {
     "model_module": "@jupyter-widgets/controls",
     "model_name": "FloatProgressModel",
     "state": {
      "_dom_classes": [],
      "_model_module": "@jupyter-widgets/controls",
      "_model_module_version": "1.5.0",
      "_model_name": "FloatProgressModel",
      "_view_count": null,
      "_view_module": "@jupyter-widgets/controls",
      "_view_module_version": "1.5.0",
      "_view_name": "ProgressView",
      "bar_style": "success",
      "description": "",
      "description_tooltip": null,
      "layout": "IPY_MODEL_be7f1c9303f44b00bccfed619c1d2215",
      "max": 1,
      "min": 0,
      "orientation": "horizontal",
      "style": "IPY_MODEL_9391e08a692e44cdbf3222208fe373cb",
      "value": 1
     }
    }
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}