01_data_import.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "4c4452b4-2520-4748-bd6d-2ae484e70b16",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Modify the file 00_setup.py to define input/output file paths on your system\n",
    "# The information in 00_setup.py will be used across notebooks\n",
    "from importlib.machinery import SourceFileLoader\n",
    "setup = SourceFileLoader(\"setup\", \"./00_setup.py\").load_module()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a595d38b-794b-4317-b39f-58da027df783",
   "metadata": {},
   "source": [
    "# 01:  Data Import\n",
    " * Import data\n",
    " * Filter targets/ nontargets\n",
    " * One hot encoding and other pre-processing\n",
    " * Split"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f5cfeb88-9b8e-44e6-91ba-61b54b6fca52",
   "metadata": {},
   "source": [
    "I do not do extensive data exploration but instead use others' insights from various\n",
    "Kaggle projects, including:\n",
    "* https://www.kaggle.com/code/mariiagusarova/feature-selection-techniques\n",
    "* https://www.kaggle.com/datasets/wordsforthewise/lending-club/\n",
    "* https://www.kaggle.com/code/wordsforthewise/eda-with-python/notebook"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6e6da1d7-27bc-40dc-bf59-4062c143c508",
   "metadata": {},
   "source": [
    "*This script takes about 1 minute on my MacBook Air*"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "5f2f4da8-290e-4287-ba3c-eef1dfb093f8",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import re\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "71df4185-bc4e-404b-a2df-6e58259b2381",
   "metadata": {},
   "outputs": [],
   "source": [
    "from pathlib import Path\n",
    "from sklearn.model_selection import train_test_split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "d46e519c-bce9-43e0-b5ee-3708982b760f",
   "metadata": {},
   "outputs": [],
   "source": [
    "from interactions_package.interactions_package import ip_plot\n",
    "ip_plot.plot_defaults()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "55ec3482-660d-4f05-9f80-2086668d6768",
   "metadata": {},
   "source": [
    "## Import Data and Filter"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "92bd280f-377e-4e47-8f83-331c8eca3768",
   "metadata": {},
   "outputs": [],
   "source": [
    "loans_init = pd.read_csv(setup.input_path,  dtype=setup.input_dtypes,\n",
    "                        parse_dates = setup.input_dates)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "21988b13-ca41-43d9-9994-b5afad2245a3",
   "metadata": {},
   "source": [
    "##### Filter to include only paid or unpaid loans"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "33474607-9463-4168-a191-2067bc910a99",
   "metadata": {},
   "outputs": [],
   "source": [
    "loans_init = loans_init[loans_init['loan_status'].isin(['Fully Paid',  'Charged Off'])]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "b84f90e7-d340-4175-8268-a8ac8cb080b5",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1345310, 151)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "loans_init.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "148af406-beb1-41b9-91ef-4286d1ed0c97",
   "metadata": {},
   "source": [
    "##### Create the target feature (0/1) that indicates a default"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "bafa5a82-2ef2-4e89-9f64-2c5c3fe31e2c",
   "metadata": {},
   "outputs": [],
   "source": [
    "loans_init['target'] = np.where(loans_init['loan_status'] == 'Fully Paid', 0, 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "42677d14-e9ae-45dc-82c8-6e3e0862fcce",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    0.800374\n",
       "1    0.199626\n",
       "Name: target, dtype: float64"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "loans_init['target'].value_counts(dropna=False, normalize=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e592f0be-9d5e-45d7-a109-f14c08f8a07c",
   "metadata": {},
   "source": [
    "##### Save full dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "678a7cde-1b80-4721-aae8-45399ea56fbc",
   "metadata": {},
   "outputs": [],
   "source": [
    "loans_init.to_parquet(Path(setup.temp_path).joinpath('01_DATA_import_filtered.parquet'))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9b788eed-2c90-461c-8783-a25ec97a5418",
   "metadata": {},
   "source": [
    "## FICO Features - Plots\n",
    "I want to determine which FICO features to include, if any"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "34510dfd-c25c-4efc-8d42-28f7f38dd5fe",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>fico_range_low</th>\n",
       "      <th>fico_range_high</th>\n",
       "      <th>last_fico_range_low</th>\n",
       "      <th>last_fico_range_high</th>\n",
       "      <th>sec_app_fico_range_low</th>\n",
       "      <th>sec_app_fico_range_high</th>\n",
       "      <th>target</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>fico_range_low</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.246554</td>\n",
       "      <td>0.344367</td>\n",
       "      <td>0.418415</td>\n",
       "      <td>0.418415</td>\n",
       "      <td>-0.130683</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fico_range_high</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.246554</td>\n",
       "      <td>0.344367</td>\n",
       "      <td>0.418419</td>\n",
       "      <td>0.418419</td>\n",
       "      <td>-0.130682</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_fico_range_low</th>\n",
       "      <td>0.246554</td>\n",
       "      <td>0.246554</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.829519</td>\n",
       "      <td>0.232893</td>\n",
       "      <td>0.232893</td>\n",
       "      <td>-0.575706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_fico_range_high</th>\n",
       "      <td>0.344367</td>\n",
       "      <td>0.344367</td>\n",
       "      <td>0.829519</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.321321</td>\n",
       "      <td>0.321321</td>\n",
       "      <td>-0.667842</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sec_app_fico_range_low</th>\n",
       "      <td>0.418415</td>\n",
       "      <td>0.418419</td>\n",
       "      <td>0.232893</td>\n",
       "      <td>0.321321</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.251242</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sec_app_fico_range_high</th>\n",
       "      <td>0.418415</td>\n",
       "      <td>0.418419</td>\n",
       "      <td>0.232893</td>\n",
       "      <td>0.321321</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.251242</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>target</th>\n",
       "      <td>-0.130683</td>\n",
       "      <td>-0.130682</td>\n",
       "      <td>-0.575706</td>\n",
       "      <td>-0.667842</td>\n",
       "      <td>-0.251242</td>\n",
       "      <td>-0.251242</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                         fico_range_low  fico_range_high  last_fico_range_low  \\\n",
       "fico_range_low                 1.000000         1.000000             0.246554   \n",
       "fico_range_high                1.000000         1.000000             0.246554   \n",
       "last_fico_range_low            0.246554         0.246554             1.000000   \n",
       "last_fico_range_high           0.344367         0.344367             0.829519   \n",
       "sec_app_fico_range_low         0.418415         0.418419             0.232893   \n",
       "sec_app_fico_range_high        0.418415         0.418419             0.232893   \n",
       "target                        -0.130683        -0.130682            -0.575706   \n",
       "\n",
       "                         last_fico_range_high  sec_app_fico_range_low  \\\n",
       "fico_range_low                       0.344367                0.418415   \n",
       "fico_range_high                      0.344367                0.418419   \n",
       "last_fico_range_low                  0.829519                0.232893   \n",
       "last_fico_range_high                 1.000000                0.321321   \n",
       "sec_app_fico_range_low               0.321321                1.000000   \n",
       "sec_app_fico_range_high              0.321321                1.000000   \n",
       "target                              -0.667842               -0.251242   \n",
       "\n",
       "                         sec_app_fico_range_high    target  \n",
       "fico_range_low                          0.418415 -0.130683  \n",
       "fico_range_high                         0.418419 -0.130682  \n",
       "last_fico_range_low                     0.232893 -0.575706  \n",
       "last_fico_range_high                    0.321321 -0.667842  \n",
       "sec_app_fico_range_low                  1.000000 -0.251242  \n",
       "sec_app_fico_range_high                 1.000000 -0.251242  \n",
       "target                                 -0.251242  1.000000  "
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fico_corr = loans_init[['fico_range_low', 'fico_range_high', 'last_fico_range_low', 'last_fico_range_high',\n",
    "           'sec_app_fico_range_low', 'sec_app_fico_range_high', 'target']].corr()\n",
    "fico_corr"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "425d3780-d64f-4fbc-8145-5870eb5e06f0",
   "metadata": {},
   "outputs": [],
   "source": [
    "fico_corr.to_csv(Path(setup.temp_path).joinpath('01_REPORT_fico_correlations.csv'), index=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "2e5af719-dc70-4204-aae5-b4603d8816ae",
   "metadata": {},
   "outputs": [],
   "source": [
    "def boxplot_fico(data, fico_col = 'fico_range_low'):\n",
    "    data.boxplot(column=fico_col, by='target', figsize=(4,3))\n",
    "    plt.suptitle(None)\n",
    "    plt.tight_layout()\n",
    "    plt.gcf().savefig(Path(setup.temp_path).joinpath('01_PLOT_fico_box_' + fico_col + '.png'))\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "33abfbe7-6ed4-4a20-ac0f-cd919da403ea",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 400x300 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "boxplot_fico(loans_init)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "445f01bb-545e-494c-b2b7-241cc02a4575",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 400x300 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "boxplot_fico(loans_init, 'last_fico_range_low')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "fb71283e-fd5a-4dc7-8d75-c9447e047771",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 400x300 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "boxplot_fico(loans_init, 'sec_app_fico_range_low')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "2b652deb-dc14-40cb-8869-95a73c28f006",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr:last-of-type th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th colspan=\"4\" halign=\"left\">fico_range_low</th>\n",
       "      <th colspan=\"4\" halign=\"left\">sec_app_fico_range_low</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>min</th>\n",
       "      <th>max</th>\n",
       "      <th>median</th>\n",
       "      <th>mean</th>\n",
       "      <th>min</th>\n",
       "      <th>max</th>\n",
       "      <th>median</th>\n",
       "      <th>mean</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>application_type</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Individual</th>\n",
       "      <td>625.0</td>\n",
       "      <td>845.0</td>\n",
       "      <td>690.0</td>\n",
       "      <td>696.026669</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Joint App</th>\n",
       "      <td>660.0</td>\n",
       "      <td>845.0</td>\n",
       "      <td>695.0</td>\n",
       "      <td>704.284690</td>\n",
       "      <td>540.0</td>\n",
       "      <td>840.0</td>\n",
       "      <td>665.0</td>\n",
       "      <td>665.970209</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 fico_range_low                            \\\n",
       "                            min    max median        mean   \n",
       "application_type                                            \n",
       "Individual                625.0  845.0  690.0  696.026669   \n",
       "Joint App                 660.0  845.0  695.0  704.284690   \n",
       "\n",
       "                 sec_app_fico_range_low                            \n",
       "                                    min    max median        mean  \n",
       "application_type                                                   \n",
       "Individual                          NaN    NaN    NaN         NaN  \n",
       "Joint App                         540.0  840.0  665.0  665.970209  "
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "loans_init.groupby('application_type')[['fico_range_low', 'sec_app_fico_range_low']] \\\n",
    "    .agg(['min', 'max', 'median', 'mean'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "f7f12edd-9599-445f-93e0-3294f28a1c45",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x28dd93af0>]"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 400x300 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Scatter plot with x=y line\n",
    "# https://stackoverflow.com/questions/25497402\n",
    "loans_init[loans_init['application_type'] == 'Joint App'] \\\n",
    "    [['fico_range_low', 'sec_app_fico_range_low']] \\\n",
    "    .sample(1000) \\\n",
    "    .plot(x='sec_app_fico_range_low', y='fico_range_low', \n",
    "          kind='scatter', figsize=(4,3))\n",
    "lims = [\n",
    "    np.min([plt.gca().get_xlim(),plt.gca().get_ylim()]),\n",
    "    np.max([plt.gca().get_xlim(), plt.gca().get_ylim()]) \n",
    "]\n",
    "plt.gca().set_xlim(lims)\n",
    "plt.gca().set_ylim(lims)\n",
    "plt.gca().plot(lims, lims, 'k-', alpha=0.75, zorder=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5208cd66-19df-4c6a-a3e7-4d6857f7794c",
   "metadata": {},
   "source": [
    "Takeaways:\n",
    "  * I don't need to use both the high and low range features, just low should be enough\n",
    "  * There is a correlation between default and the second app's fico (for those with second applications), which may be of interest. \n",
    "  * Most of the time, the second applicant has the lower interest rate, often much lower."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "50d49d06-2d56-4a35-95b8-d33bd4fa0110",
   "metadata": {},
   "source": [
    "## Special Processing of Some Features\n",
    "I plan to use a tree-based model and so I am OK with ordinal or thresholded values"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "83ad33a5-72a3-4bf9-99cf-d602e8307c26",
   "metadata": {},
   "source": [
    "##### Term - extract number of months"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "0031c1f6-5226-490c-bb82-76792ad0d298",
   "metadata": {},
   "outputs": [],
   "source": [
    "def term_extract(instr):\n",
    "    try:\n",
    "        time = instr.split()[1].lower()\n",
    "    except:\n",
    "        return None\n",
    "    \n",
    "    if 'month' in time:\n",
    "        return int(instr.split()[0])\n",
    "    elif 'year' in time:\n",
    "        return int(instr.split()[0]) * 12\n",
    "    \n",
    "    return None"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "5a087f70-4ffc-4ae3-916c-cf0f131d677f",
   "metadata": {},
   "outputs": [],
   "source": [
    "loans_init['term'] = loans_init['term'].apply(term_extract)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1b3e6319-8935-4d6f-8e60-27ae8d3a5a45",
   "metadata": {},
   "source": [
    "##### Employment Length - extract years"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "b2113bbb-9dea-4c54-8652-bcfed874907c",
   "metadata": {},
   "outputs": [],
   "source": [
    "def emp_length_extract(instr):\n",
    "    if instr == '< 1 year':\n",
    "        return 0\n",
    "    elif instr == '10+ years':\n",
    "        return 10\n",
    "    \n",
    "    try:\n",
    "        time = instr.split()[1].lower()\n",
    "    except:\n",
    "        return None\n",
    "    \n",
    "    if 'year' in time:\n",
    "        return int(instr.split()[0]) \n",
    "    \n",
    "    return None"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "5430e931-deca-451c-b6de-9af26d1cd2f2",
   "metadata": {},
   "outputs": [],
   "source": [
    "loans_init['emp_length'] = loans_init['emp_length'].apply(emp_length_extract)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d881ccc9-0ca9-4496-b89c-f4aefe7f30e2",
   "metadata": {},
   "source": [
    "##### (Sub) Grade - Use float value to encode an ordinal rating \n",
    "I do not use this feature in the final model as its meaning is proprietory and it is extremely predictive."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "b27e5d5b-3c37-4e7c-b9f2-3ddecd48a0b8",
   "metadata": {},
   "outputs": [],
   "source": [
    "def ord_grade(grade_str):\n",
    "    \"\"\" Encode the detailed grade.  Smaller values are better.\n",
    "    Take the letter grade as a whole number, using its ASCII code\n",
    "    relative to \"A\".  Get sub-grades as decimals.\"\"\"\n",
    "    \n",
    "    # Get the whole part of the encoded value\n",
    "    try:\n",
    "        grade_base = ord(grade_str[0].upper()) - ord('A')\n",
    "    except:\n",
    "        return None\n",
    "    \n",
    "    # Get the decimal portion\n",
    "    try:\n",
    "        grade_dec = np.min([(int(grade_str[1]) - 1) * 0.2, 0.8])\n",
    "    except:\n",
    "        return grade_base\n",
    "    \n",
    "    return grade_base + grade_dec"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "dfa99124-ac43-4e40-8655-60aafc2654eb",
   "metadata": {},
   "outputs": [],
   "source": [
    "loans_init['sub_grade'] = loans_init['sub_grade'].apply(ord_grade)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "344310f3-0a38-45ba-9d32-e8507a7b511f",
   "metadata": {},
   "source": [
    "##### Verification status - make binary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "8fe1915a-aabb-443c-8135-49acab52645e",
   "metadata": {},
   "outputs": [],
   "source": [
    "loans_init['verification_status'] = np.where(loans_init['verification_status'] \\\n",
    "                                                .isin(['Source Verified', 'Verified']),\n",
    "                                             1, 0)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9a95a731-49c1-49d3-a3f5-8c6b96441f28",
   "metadata": {},
   "source": [
    "##### Application type - make binary\n",
    "I wind up not using this feature in the model, either, as it does not contribute to the model if we include the second application FICO."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "5d9e0bd1-9ee7-4991-a09c-b305412af089",
   "metadata": {},
   "outputs": [],
   "source": [
    "loans_init['application_type'] = np.where(loans_init['application_type'] \\\n",
    "                                                 .str.lower().str.contains('joint'),\n",
    "                                          1, 0)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ec3813f9-4665-43ee-b389-3b38ce791ac8",
   "metadata": {},
   "source": [
    "##### Initial list status - make binary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "20f1492d-8ff0-464f-8c33-64c3ef737238",
   "metadata": {},
   "outputs": [],
   "source": [
    "loans_init['initial_list_status'] = np.where(loans_init['initial_list_status'] =='w', 1, 0)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "674765cd-6a6b-46fa-a6d3-cb9a52ca33fb",
   "metadata": {},
   "source": [
    "## One Hot Encode Remaining Strings"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "898900b5-3b0f-4ae8-8e2d-d869c0f12b44",
   "metadata": {},
   "source": [
    "##### Function to one-hot encode features, keeping the top categories\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "bad06f45-4ead-47cf-91d4-c57f00414592",
   "metadata": {},
   "outputs": [],
   "source": [
    "def one_hot_encode_thresh(ser, num_levels_keep = 3,\n",
    "                          min_pct = 0.01,\n",
    "                          other_level_name = 'ZZ_OTHER',\n",
    "                          prefix_sep = '__'):\n",
    "    \"\"\" One hot encode a feature, keeping only the most commmon levels,\n",
    "    determined by a max number and percent threshold.  Any other categories \n",
    "    will be put into an \"other\" bucket.  The level name strings are \n",
    "    normalized,so they contain only alphanumerics and underscores.  Level\n",
    "    names also capitalized.\n",
    "      Inputs:\n",
    "          ser:  The series to one hot encode\n",
    "          num_levels_keep: Number of levels to retain\n",
    "          min_pct: Min percent of population to retain a level,\n",
    "              regardless of num_levels_keep \n",
    "          other_level_name: String used to label the other \n",
    "              category, if applicable.\n",
    "          prefix_sep:  Separator for the one-hot encoding\n",
    "          prefix:  One-hot encoding prefix\n",
    "      Value:  A data frame containing the one-hot encoded values\"\"\"\n",
    "    \n",
    "    # Get the most common categories\n",
    "    all_value_counts = ser.value_counts(normalize=True)\n",
    "    all_levels = all_value_counts.index.tolist()\n",
    "    top_feat = all_value_counts[all_value_counts >= min_pct] \\\n",
    "        .index.tolist()[0:num_levels_keep]\n",
    "\n",
    "    # Get the levels to keep, up to the max\n",
    "    # If we have only one level in \"other\", don't renane it\n",
    "    if len(all_levels) > len(top_feat) + 1:\n",
    "        top_feat.append(other_level_name)\n",
    "    elif len(all_levels) == len(top_feat) + 1:\n",
    "        top_feat = all_levels\n",
    "        \n",
    "    # Make feature categorical\n",
    "    cat_feat = pd.Categorical(ser, categories=top_feat)\n",
    "    \n",
    "    # Normalize category strings, if necessary\n",
    "    cat_rename_dict = {k: re.sub('[^0-9a-zA-Z_]+', '', k).upper() for k in cat_feat.categories}\n",
    "    cat_feat = cat_feat.rename_categories(cat_rename_dict)\n",
    "    \n",
    "    # One hot encode\n",
    "    return pd.get_dummies(cat_feat, prefix = ser.name, prefix_sep = prefix_sep)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fb4d3ae0-8635-46ec-b87d-9c9795345a55",
   "metadata": {},
   "source": [
    "##### Encode remaining string features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "9081015b-ed50-4dfc-a0ee-c1326f89fbf8",
   "metadata": {},
   "outputs": [],
   "source": [
    "one_hot_feat = set(loans_init[setup.predictor_features].select_dtypes(include=['object']).columns) - \\\n",
    "    set(['term', 'emp_length', 'sub_grade', 'verification_status', 'application_type', 'initial_list_status'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "5dd6098d-844d-4eec-8984-9ebac7a08602",
   "metadata": {},
   "outputs": [],
   "source": [
    "loans_onehot = pd.concat([one_hot_encode_thresh(loans_init[d]) for d in one_hot_feat],\n",
    "                         axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "f0e941bf-00f4-465c-bf67-2c2e257ec621",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1345310, 4)"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "loans_onehot.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "52a92dac-fed8-416b-bc30-ed8c8c4b121a",
   "metadata": {},
   "source": [
    "##### Save binary features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "e11345ab-aad1-43f6-8b9c-54e9687db0f0",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['home_ownership__MORTGAGE', 'home_ownership__RENT', 'home_ownership__OWN', 'home_ownership__ZZ_OTHER', 'verification_status', 'initial_list_status']\n"
     ]
    }
   ],
   "source": [
    "binary_features = list(loans_onehot.columns) + \\\n",
    "    [f for f in ['application_type', 'verification_status', 'initial_list_status'] \\\n",
    "     if f in setup.predictor_features]\n",
    "print(binary_features)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "3b98cab7-5636-4779-a466-c4b0446984dd",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save the binary list features\n",
    "pd.DataFrame({'feature':binary_features}) \\\n",
    "    .drop_duplicates() \\\n",
    "    .to_csv(Path(setup.temp_path).joinpath('01_REPORT_predictor_features_binary.csv'))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e3a4a047-4ff4-4d77-9c3d-93f1fab5aa92",
   "metadata": {},
   "source": [
    "##### Finalize dataset, keeping only selected features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "4e01751d-d3cb-4d6c-ba72-f9b18e88b10d",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Keep only selected features\n",
    "used=set()\n",
    "feat_list = [f for f in setup.info_features + setup.predictor_features + ['target'] \\\n",
    "             if f in loans_init.columns and f not in used \\\n",
    "             and f not in loans_onehot and (used.add(f) or True)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "270c1a53-bf9d-4cbe-b844-65014d8e1ce1",
   "metadata": {},
   "outputs": [],
   "source": [
    "loans_fin = pd.concat([loans_init[feat_list].reset_index(), loans_onehot], \n",
    "                      axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "1b0e70e8-3a6b-4ddb-a0b6-b6bcd7d638bf",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1345310, 35)"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "loans_fin.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "9543ff5d-c62a-4239-939b-8254421d59e5",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>index</th>\n",
       "      <th>id</th>\n",
       "      <th>grade</th>\n",
       "      <th>sub_grade</th>\n",
       "      <th>hardship_flag</th>\n",
       "      <th>debt_settlement_flag</th>\n",
       "      <th>hardship_amount</th>\n",
       "      <th>settlement_amount</th>\n",
       "      <th>addr_state</th>\n",
       "      <th>purpose</th>\n",
       "      <th>...</th>\n",
       "      <th>initial_list_status</th>\n",
       "      <th>num_actv_bc_tl</th>\n",
       "      <th>mort_acc</th>\n",
       "      <th>pub_rec</th>\n",
       "      <th>revol_bal</th>\n",
       "      <th>target</th>\n",
       "      <th>home_ownership__MORTGAGE</th>\n",
       "      <th>home_ownership__RENT</th>\n",
       "      <th>home_ownership__OWN</th>\n",
       "      <th>home_ownership__ZZ_OTHER</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>68407277</td>\n",
       "      <td>C</td>\n",
       "      <td>2.6</td>\n",
       "      <td>N</td>\n",
       "      <td>N</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>PA</td>\n",
       "      <td>debt_consolidation</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2765.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>68355089</td>\n",
       "      <td>C</td>\n",
       "      <td>2.0</td>\n",
       "      <td>N</td>\n",
       "      <td>N</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>SD</td>\n",
       "      <td>small_business</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>5.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>21470.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>68341763</td>\n",
       "      <td>B</td>\n",
       "      <td>1.6</td>\n",
       "      <td>N</td>\n",
       "      <td>N</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>IL</td>\n",
       "      <td>home_improvement</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>2.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>7869.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>3 rows × 35 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   index        id grade  sub_grade hardship_flag debt_settlement_flag  \\\n",
       "0      0  68407277     C        2.6             N                    N   \n",
       "1      1  68355089     C        2.0             N                    N   \n",
       "2      2  68341763     B        1.6             N                    N   \n",
       "\n",
       "   hardship_amount  settlement_amount addr_state             purpose  ...  \\\n",
       "0              NaN                NaN         PA  debt_consolidation  ...   \n",
       "1              NaN                NaN         SD      small_business  ...   \n",
       "2              NaN                NaN         IL    home_improvement  ...   \n",
       "\n",
       "   initial_list_status  num_actv_bc_tl  mort_acc  pub_rec  revol_bal  target  \\\n",
       "0                    1             2.0       1.0      0.0     2765.0       0   \n",
       "1                    1             5.0       4.0      0.0    21470.0       0   \n",
       "2                    1             2.0       5.0      0.0     7869.0       0   \n",
       "\n",
       "   home_ownership__MORTGAGE  home_ownership__RENT  home_ownership__OWN  \\\n",
       "0                         1                     0                    0   \n",
       "1                         1                     0                    0   \n",
       "2                         1                     0                    0   \n",
       "\n",
       "  home_ownership__ZZ_OTHER  \n",
       "0                        0  \n",
       "1                        0  \n",
       "2                        0  \n",
       "\n",
       "[3 rows x 35 columns]"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "loans_fin.head(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e77e696c-e52c-49f0-b2bc-c4f42535590a",
   "metadata": {},
   "source": [
    "##### Save predictor feature list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "b9da7575-abbd-40a4-85db-f49b14df0b39",
   "metadata": {},
   "outputs": [],
   "source": [
    "predictor_features =  [f for f in setup.predictor_features + list(loans_onehot.columns) \\\n",
    "                       if f in loans_fin.select_dtypes(include=np.number).columns]\n",
    "pd.DataFrame({'feature':predictor_features}) \\\n",
    "    .drop_duplicates() \\\n",
    "    .to_csv(Path(setup.temp_path).joinpath('01_REPORT_predictor_features.csv'), index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "f41b9b2e-b83b-4e41-8942-b6322ee6f3a2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "18"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(predictor_features)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "7ed1e6a8-d643-4ba3-bbc5-6767a2a7410a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['loan_amnt', 'term', 'int_rate', 'emp_length', 'annual_inc', 'verification_status', 'fico_range_low', 'sec_app_fico_range_low', 'open_acc', 'initial_list_status', 'num_actv_bc_tl', 'mort_acc', 'pub_rec', 'revol_bal', 'home_ownership__MORTGAGE', 'home_ownership__RENT', 'home_ownership__OWN', 'home_ownership__ZZ_OTHER']\n"
     ]
    }
   ],
   "source": [
    "print(predictor_features)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "76132611-7887-40ef-9341-9485621aa2b6",
   "metadata": {},
   "source": [
    "## NA Fill\n",
    "If I were doing an xgboost model, I could skip this step, but I will do a GradientBoostingClassifier instead, as this model type is compatible with the package to calculate Friedman's H (sklearn_gbmi)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "b9c5201d-222f-461e-b16e-a615d8dce92b",
   "metadata": {},
   "outputs": [],
   "source": [
    "na_dset = loans_fin[setup.predictor_features]\\\n",
    "    .select_dtypes('number') \\\n",
    "    .apply(np.isnan) \\\n",
    "    .apply(pd.Series.value_counts)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "514bf6a9-554c-4398-87af-3ee906a5da8d",
   "metadata": {},
   "outputs": [],
   "source": [
    "na_dset_T = na_dset.T.reset_index()\n",
    "na_features = na_dset_T[~na_dset_T[True].isna()]['index']"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3754f9bc-e999-4aa4-9d46-d5cf68937f5a",
   "metadata": {},
   "source": [
    "##### Second application FICO\n",
    "Use first app score when missing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "56c2cf98-614f-4f99-b4ae-17b53c56c378",
   "metadata": {},
   "outputs": [],
   "source": [
    "loans_fin['sec_app_fico_range_low'] = loans_fin['sec_app_fico_range_low'] \\\n",
    "    .where(~loans_fin['sec_app_fico_range_low'].isna(),\n",
    "           loans_fin['fico_range_low'])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1af25ed2-3855-4fba-83b8-dd9916914396",
   "metadata": {},
   "source": [
    "##### Features to set to 0 (counters)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "02f70746-0b2f-4219-b442-cfbff661c3ed",
   "metadata": {},
   "outputs": [],
   "source": [
    "counter_nas = [c for c in na_features if \n",
    "               c.endswith('acc') or c.startswith('num') or \n",
    "               c.endswith('acc') or c.startswith('inq') or \n",
    "               c.endswith('util') or\n",
    "               c in ['emp_length']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "1acc53b0-d7f6-447a-9059-cfc7cfaaf9bb",
   "metadata": {},
   "outputs": [],
   "source": [
    "loans_fin[counter_nas] = loans_fin[counter_nas].fillna(0)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "83f4b4f9-c53e-4a22-b2dd-e5c90a847106",
   "metadata": {},
   "source": [
    "##### Use median for the rest (if any)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "5d52d326-90d8-4278-a2a4-c20dde8658db",
   "metadata": {},
   "outputs": [],
   "source": [
    "median_feat = [c for c in na_features if \n",
    "               not (c in counter_nas + ['sec_app_fico_range_low'])]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "56164236-cc6d-4184-b396-49871be867f7",
   "metadata": {},
   "outputs": [],
   "source": [
    "if len(median_feat) > 0:\n",
    "    median_val_dict = loans_fin[median_feat].apply(np.nanmedian).to_dict()\n",
    "    loans_fin[median_feat] = loans_fin[median_feat] \\\n",
    "        .apply(lambda x: x.fillna(median_val_dict[x.name]))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c6f78387-64dd-4fe7-9bfc-ebbd4e6150f0",
   "metadata": {},
   "source": [
    "## Split the data\n",
    " * Do a 65/20/15 split"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "952e7677-d268-46fb-aab4-18ab85320b90",
   "metadata": {},
   "source": [
    "##### Create hold-out"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "580b130c-9f37-4e12-975b-35bc30e73bfd",
   "metadata": {},
   "source": [
    "##### Do train-test-validation split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "e82b5d51-c232-4b0a-9b8b-e9411abca751",
   "metadata": {},
   "outputs": [],
   "source": [
    "[base_train, base_other] = train_test_split(loans_fin, train_size=0.65, random_state=1225,\n",
    "                                                    stratify=loans_fin['target'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "8b335dbd-6e35-4fe1-bd7e-c38835cfbfb0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(874451, 35)"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "base_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "17b1748b-8759-413d-b138-b1281098f295",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(470859, 35)"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "base_other.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "bd5d1d5d-81cf-427a-9f1e-e1bdb489ce3b",
   "metadata": {},
   "outputs": [],
   "source": [
    "base_train.to_parquet(Path(setup.temp_path).joinpath('01_DATA_combined_base_train.parquet'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "83c4898b-910a-4f1c-8d98-5d1ef6fb5f90",
   "metadata": {},
   "outputs": [],
   "source": [
    "[base_test, base_val] = train_test_split(base_other, train_size=0.8, random_state=123,\n",
    "                                                    stratify=base_other['target'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "f1bfc30f-c9dc-41a8-997d-a31d80f211b5",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(376687, 35)"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "base_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "f005c210-f7f8-4d0c-acdb-640d82315a02",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(94172, 35)"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "base_val.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "75040f22-417f-49f3-aad0-4cf5d7c8f405",
   "metadata": {},
   "outputs": [],
   "source": [
    "base_test.to_parquet(Path(setup.temp_path).joinpath('01_DATA_combined_base_test.parquet'))\n",
    "base_val.to_parquet(Path(setup.temp_path).joinpath('01_DATA_combined_base_val.parquet'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b47a4381-21ff-4f77-aa55-9575f5d3efe0",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.10.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}