{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "iraqi-wound",
   "metadata": {},
   "source": [
    "## Dependencies"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "legal-router",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: imbalanced-learn in /opt/conda/lib/python3.8/site-packages (0.8.0)\n",
      "Requirement already satisfied: scipy>=0.19.1 in /opt/conda/lib/python3.8/site-packages (from imbalanced-learn) (1.6.0)\n",
      "Requirement already satisfied: joblib>=0.11 in /opt/conda/lib/python3.8/site-packages (from imbalanced-learn) (1.0.0)\n",
      "Requirement already satisfied: numpy>=1.13.3 in /opt/conda/lib/python3.8/site-packages (from imbalanced-learn) (1.20.0)\n",
      "Requirement already satisfied: scikit-learn>=0.24 in /opt/conda/lib/python3.8/site-packages (from imbalanced-learn) (0.24.1)\n",
      "Requirement already satisfied: threadpoolctl>=2.0.0 in /opt/conda/lib/python3.8/site-packages (from scikit-learn>=0.24->imbalanced-learn) (2.1.0)\n",
      "Requirement already satisfied: python-slugify in /opt/conda/lib/python3.8/site-packages (4.0.1)\n",
      "Requirement already satisfied: text-unidecode>=1.3 in /opt/conda/lib/python3.8/site-packages (from python-slugify) (1.3)\n"
     ]
    }
   ],
   "source": [
    "import sys\n",
    "!{sys.executable} -m pip install -U imbalanced-learn\n",
    "!{sys.executable} -m pip install -U python-slugify"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "superior-clinton",
   "metadata": {},
   "source": [
    "## Dataset overview"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "assumed-progress",
   "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>Bankrupt?</th>\n",
       "      <th>ROA(C) before interest and depreciation before interest</th>\n",
       "      <th>ROA(A) before interest and % after tax</th>\n",
       "      <th>ROA(B) before interest and depreciation after tax</th>\n",
       "      <th>Operating Gross Margin</th>\n",
       "      <th>Realized Sales Gross Margin</th>\n",
       "      <th>Operating Profit Rate</th>\n",
       "      <th>Pre-tax net Interest Rate</th>\n",
       "      <th>After-tax net Interest Rate</th>\n",
       "      <th>Non-industry income and expenditure/revenue</th>\n",
       "      <th>...</th>\n",
       "      <th>Net Income to Total Assets</th>\n",
       "      <th>Total assets to GNP price</th>\n",
       "      <th>No-credit Interval</th>\n",
       "      <th>Gross Profit to Sales</th>\n",
       "      <th>Net Income to Stockholder's Equity</th>\n",
       "      <th>Liability to Equity</th>\n",
       "      <th>Degree of Financial Leverage (DFL)</th>\n",
       "      <th>Interest Coverage Ratio (Interest expense to EBIT)</th>\n",
       "      <th>Net Income Flag</th>\n",
       "      <th>Equity to Liability</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>0.370594</td>\n",
       "      <td>0.424389</td>\n",
       "      <td>0.405750</td>\n",
       "      <td>0.601457</td>\n",
       "      <td>0.601457</td>\n",
       "      <td>0.998969</td>\n",
       "      <td>0.796887</td>\n",
       "      <td>0.808809</td>\n",
       "      <td>0.302646</td>\n",
       "      <td>...</td>\n",
       "      <td>0.716845</td>\n",
       "      <td>0.009219</td>\n",
       "      <td>0.622879</td>\n",
       "      <td>0.601453</td>\n",
       "      <td>0.827890</td>\n",
       "      <td>0.290202</td>\n",
       "      <td>0.026601</td>\n",
       "      <td>0.564050</td>\n",
       "      <td>1</td>\n",
       "      <td>0.016469</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>0.464291</td>\n",
       "      <td>0.538214</td>\n",
       "      <td>0.516730</td>\n",
       "      <td>0.610235</td>\n",
       "      <td>0.610235</td>\n",
       "      <td>0.998946</td>\n",
       "      <td>0.797380</td>\n",
       "      <td>0.809301</td>\n",
       "      <td>0.303556</td>\n",
       "      <td>...</td>\n",
       "      <td>0.795297</td>\n",
       "      <td>0.008323</td>\n",
       "      <td>0.623652</td>\n",
       "      <td>0.610237</td>\n",
       "      <td>0.839969</td>\n",
       "      <td>0.283846</td>\n",
       "      <td>0.264577</td>\n",
       "      <td>0.570175</td>\n",
       "      <td>1</td>\n",
       "      <td>0.020794</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>0.426071</td>\n",
       "      <td>0.499019</td>\n",
       "      <td>0.472295</td>\n",
       "      <td>0.601450</td>\n",
       "      <td>0.601364</td>\n",
       "      <td>0.998857</td>\n",
       "      <td>0.796403</td>\n",
       "      <td>0.808388</td>\n",
       "      <td>0.302035</td>\n",
       "      <td>...</td>\n",
       "      <td>0.774670</td>\n",
       "      <td>0.040003</td>\n",
       "      <td>0.623841</td>\n",
       "      <td>0.601449</td>\n",
       "      <td>0.836774</td>\n",
       "      <td>0.290189</td>\n",
       "      <td>0.026555</td>\n",
       "      <td>0.563706</td>\n",
       "      <td>1</td>\n",
       "      <td>0.016474</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>0.399844</td>\n",
       "      <td>0.451265</td>\n",
       "      <td>0.457733</td>\n",
       "      <td>0.583541</td>\n",
       "      <td>0.583541</td>\n",
       "      <td>0.998700</td>\n",
       "      <td>0.796967</td>\n",
       "      <td>0.808966</td>\n",
       "      <td>0.303350</td>\n",
       "      <td>...</td>\n",
       "      <td>0.739555</td>\n",
       "      <td>0.003252</td>\n",
       "      <td>0.622929</td>\n",
       "      <td>0.583538</td>\n",
       "      <td>0.834697</td>\n",
       "      <td>0.281721</td>\n",
       "      <td>0.026697</td>\n",
       "      <td>0.564663</td>\n",
       "      <td>1</td>\n",
       "      <td>0.023982</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>0.465022</td>\n",
       "      <td>0.538432</td>\n",
       "      <td>0.522298</td>\n",
       "      <td>0.598783</td>\n",
       "      <td>0.598783</td>\n",
       "      <td>0.998973</td>\n",
       "      <td>0.797366</td>\n",
       "      <td>0.809304</td>\n",
       "      <td>0.303475</td>\n",
       "      <td>...</td>\n",
       "      <td>0.795016</td>\n",
       "      <td>0.003878</td>\n",
       "      <td>0.623521</td>\n",
       "      <td>0.598782</td>\n",
       "      <td>0.839973</td>\n",
       "      <td>0.278514</td>\n",
       "      <td>0.024752</td>\n",
       "      <td>0.575617</td>\n",
       "      <td>1</td>\n",
       "      <td>0.035490</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 96 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   Bankrupt?   ROA(C) before interest and depreciation before interest  \\\n",
       "0          1                                           0.370594          \n",
       "1          1                                           0.464291          \n",
       "2          1                                           0.426071          \n",
       "3          1                                           0.399844          \n",
       "4          1                                           0.465022          \n",
       "\n",
       "    ROA(A) before interest and % after tax  \\\n",
       "0                                 0.424389   \n",
       "1                                 0.538214   \n",
       "2                                 0.499019   \n",
       "3                                 0.451265   \n",
       "4                                 0.538432   \n",
       "\n",
       "    ROA(B) before interest and depreciation after tax  \\\n",
       "0                                           0.405750    \n",
       "1                                           0.516730    \n",
       "2                                           0.472295    \n",
       "3                                           0.457733    \n",
       "4                                           0.522298    \n",
       "\n",
       "    Operating Gross Margin   Realized Sales Gross Margin  \\\n",
       "0                 0.601457                      0.601457   \n",
       "1                 0.610235                      0.610235   \n",
       "2                 0.601450                      0.601364   \n",
       "3                 0.583541                      0.583541   \n",
       "4                 0.598783                      0.598783   \n",
       "\n",
       "    Operating Profit Rate   Pre-tax net Interest Rate  \\\n",
       "0                0.998969                    0.796887   \n",
       "1                0.998946                    0.797380   \n",
       "2                0.998857                    0.796403   \n",
       "3                0.998700                    0.796967   \n",
       "4                0.998973                    0.797366   \n",
       "\n",
       "    After-tax net Interest Rate   Non-industry income and expenditure/revenue  \\\n",
       "0                      0.808809                                      0.302646   \n",
       "1                      0.809301                                      0.303556   \n",
       "2                      0.808388                                      0.302035   \n",
       "3                      0.808966                                      0.303350   \n",
       "4                      0.809304                                      0.303475   \n",
       "\n",
       "   ...   Net Income to Total Assets   Total assets to GNP price  \\\n",
       "0  ...                     0.716845                    0.009219   \n",
       "1  ...                     0.795297                    0.008323   \n",
       "2  ...                     0.774670                    0.040003   \n",
       "3  ...                     0.739555                    0.003252   \n",
       "4  ...                     0.795016                    0.003878   \n",
       "\n",
       "    No-credit Interval   Gross Profit to Sales  \\\n",
       "0             0.622879                0.601453   \n",
       "1             0.623652                0.610237   \n",
       "2             0.623841                0.601449   \n",
       "3             0.622929                0.583538   \n",
       "4             0.623521                0.598782   \n",
       "\n",
       "    Net Income to Stockholder's Equity   Liability to Equity  \\\n",
       "0                             0.827890              0.290202   \n",
       "1                             0.839969              0.283846   \n",
       "2                             0.836774              0.290189   \n",
       "3                             0.834697              0.281721   \n",
       "4                             0.839973              0.278514   \n",
       "\n",
       "    Degree of Financial Leverage (DFL)  \\\n",
       "0                             0.026601   \n",
       "1                             0.264577   \n",
       "2                             0.026555   \n",
       "3                             0.026697   \n",
       "4                             0.024752   \n",
       "\n",
       "    Interest Coverage Ratio (Interest expense to EBIT)   Net Income Flag  \\\n",
       "0                                           0.564050                   1   \n",
       "1                                           0.570175                   1   \n",
       "2                                           0.563706                   1   \n",
       "3                                           0.564663                   1   \n",
       "4                                           0.575617                   1   \n",
       "\n",
       "    Equity to Liability  \n",
       "0              0.016469  \n",
       "1              0.020794  \n",
       "2              0.016474  \n",
       "3              0.023982  \n",
       "4              0.035490  \n",
       "\n",
       "[5 rows x 96 columns]"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "\n",
    "df = pd.read_csv('company_bankrupcy.csv').dropna()\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "premier-lincoln",
   "metadata": {},
   "source": [
    "## Imbalance Analysis\n",
    "\n",
    "\n",
    "Is always helpful to know how imbalance the dataset is. Are there more positive samples than negatives ? How much ?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "cooked-april",
   "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>count</th>\n",
       "      <th>pct</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>OK</th>\n",
       "      <td>6599</td>\n",
       "      <td>0.967737</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>KO</th>\n",
       "      <td>220</td>\n",
       "      <td>0.032263</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    count       pct\n",
       "OK   6599  0.967737\n",
       "KO    220  0.032263"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "count = len(df)\n",
    "stats = df[['Bankrupt?', ' After-tax net Interest Rate']]\\\n",
    "    .groupby(['Bankrupt?'])\\\n",
    "    .agg(['count'])\\\n",
    "    .reset_index(drop=True)\\\n",
    "    .rename(index={0: \"OK\", 1: \"KO\"})\\\n",
    "    .T.reset_index(drop=True, level=0).T\n",
    "\n",
    "stats['pct'] = stats['count'] / count\n",
    "stats"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "based-virtue",
   "metadata": {},
   "source": [
    "It seems this dataset is highly imbalanced. Most of the samples are negative target samples, whereas only 3% of the samples are of type positive."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "imperial-clinton",
   "metadata": {},
   "source": [
    "### Imbalanced: Resampling & Feature selection\n",
    "\n",
    "We need to narrow down the number of features that are really relevant for classifying the samples between bankrupcy and bussiness as usual. The steps I'm going to do are:\n",
    "\n",
    "- Correlation Matrix\n",
    "- Scatter Matrix\n",
    "- Lasso feature selection\n",
    "- SBKBest feature selection"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "sitting-posting",
   "metadata": {},
   "source": [
    "First I'm normalizing the **dataframe's columns names**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "delayed-velvet",
   "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>bankrupt</th>\n",
       "      <th>roa-c-before-interest-and-depreciation-before-interest</th>\n",
       "      <th>roa-a-before-interest-and-after-tax</th>\n",
       "      <th>roa-b-before-interest-and-depreciation-after-tax</th>\n",
       "      <th>operating-gross-margin</th>\n",
       "      <th>realized-sales-gross-margin</th>\n",
       "      <th>operating-profit-rate</th>\n",
       "      <th>pre-tax-net-interest-rate</th>\n",
       "      <th>after-tax-net-interest-rate</th>\n",
       "      <th>non-industry-income-and-expenditure-revenue</th>\n",
       "      <th>...</th>\n",
       "      <th>net-income-to-total-assets</th>\n",
       "      <th>total-assets-to-gnp-price</th>\n",
       "      <th>no-credit-interval</th>\n",
       "      <th>gross-profit-to-sales</th>\n",
       "      <th>net-income-to-stockholder-s-equity</th>\n",
       "      <th>liability-to-equity</th>\n",
       "      <th>degree-of-financial-leverage-dfl</th>\n",
       "      <th>interest-coverage-ratio-interest-expense-to-ebit</th>\n",
       "      <th>net-income-flag</th>\n",
       "      <th>equity-to-liability</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>0.370594</td>\n",
       "      <td>0.424389</td>\n",
       "      <td>0.405750</td>\n",
       "      <td>0.601457</td>\n",
       "      <td>0.601457</td>\n",
       "      <td>0.998969</td>\n",
       "      <td>0.796887</td>\n",
       "      <td>0.808809</td>\n",
       "      <td>0.302646</td>\n",
       "      <td>...</td>\n",
       "      <td>0.716845</td>\n",
       "      <td>0.009219</td>\n",
       "      <td>0.622879</td>\n",
       "      <td>0.601453</td>\n",
       "      <td>0.827890</td>\n",
       "      <td>0.290202</td>\n",
       "      <td>0.026601</td>\n",
       "      <td>0.564050</td>\n",
       "      <td>1</td>\n",
       "      <td>0.016469</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>0.464291</td>\n",
       "      <td>0.538214</td>\n",
       "      <td>0.516730</td>\n",
       "      <td>0.610235</td>\n",
       "      <td>0.610235</td>\n",
       "      <td>0.998946</td>\n",
       "      <td>0.797380</td>\n",
       "      <td>0.809301</td>\n",
       "      <td>0.303556</td>\n",
       "      <td>...</td>\n",
       "      <td>0.795297</td>\n",
       "      <td>0.008323</td>\n",
       "      <td>0.623652</td>\n",
       "      <td>0.610237</td>\n",
       "      <td>0.839969</td>\n",
       "      <td>0.283846</td>\n",
       "      <td>0.264577</td>\n",
       "      <td>0.570175</td>\n",
       "      <td>1</td>\n",
       "      <td>0.020794</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>0.426071</td>\n",
       "      <td>0.499019</td>\n",
       "      <td>0.472295</td>\n",
       "      <td>0.601450</td>\n",
       "      <td>0.601364</td>\n",
       "      <td>0.998857</td>\n",
       "      <td>0.796403</td>\n",
       "      <td>0.808388</td>\n",
       "      <td>0.302035</td>\n",
       "      <td>...</td>\n",
       "      <td>0.774670</td>\n",
       "      <td>0.040003</td>\n",
       "      <td>0.623841</td>\n",
       "      <td>0.601449</td>\n",
       "      <td>0.836774</td>\n",
       "      <td>0.290189</td>\n",
       "      <td>0.026555</td>\n",
       "      <td>0.563706</td>\n",
       "      <td>1</td>\n",
       "      <td>0.016474</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>0.399844</td>\n",
       "      <td>0.451265</td>\n",
       "      <td>0.457733</td>\n",
       "      <td>0.583541</td>\n",
       "      <td>0.583541</td>\n",
       "      <td>0.998700</td>\n",
       "      <td>0.796967</td>\n",
       "      <td>0.808966</td>\n",
       "      <td>0.303350</td>\n",
       "      <td>...</td>\n",
       "      <td>0.739555</td>\n",
       "      <td>0.003252</td>\n",
       "      <td>0.622929</td>\n",
       "      <td>0.583538</td>\n",
       "      <td>0.834697</td>\n",
       "      <td>0.281721</td>\n",
       "      <td>0.026697</td>\n",
       "      <td>0.564663</td>\n",
       "      <td>1</td>\n",
       "      <td>0.023982</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>0.465022</td>\n",
       "      <td>0.538432</td>\n",
       "      <td>0.522298</td>\n",
       "      <td>0.598783</td>\n",
       "      <td>0.598783</td>\n",
       "      <td>0.998973</td>\n",
       "      <td>0.797366</td>\n",
       "      <td>0.809304</td>\n",
       "      <td>0.303475</td>\n",
       "      <td>...</td>\n",
       "      <td>0.795016</td>\n",
       "      <td>0.003878</td>\n",
       "      <td>0.623521</td>\n",
       "      <td>0.598782</td>\n",
       "      <td>0.839973</td>\n",
       "      <td>0.278514</td>\n",
       "      <td>0.024752</td>\n",
       "      <td>0.575617</td>\n",
       "      <td>1</td>\n",
       "      <td>0.035490</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 96 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   bankrupt  roa-c-before-interest-and-depreciation-before-interest  \\\n",
       "0         1                                           0.370594        \n",
       "1         1                                           0.464291        \n",
       "2         1                                           0.426071        \n",
       "3         1                                           0.399844        \n",
       "4         1                                           0.465022        \n",
       "\n",
       "   roa-a-before-interest-and-after-tax  \\\n",
       "0                             0.424389   \n",
       "1                             0.538214   \n",
       "2                             0.499019   \n",
       "3                             0.451265   \n",
       "4                             0.538432   \n",
       "\n",
       "   roa-b-before-interest-and-depreciation-after-tax  operating-gross-margin  \\\n",
       "0                                          0.405750                0.601457   \n",
       "1                                          0.516730                0.610235   \n",
       "2                                          0.472295                0.601450   \n",
       "3                                          0.457733                0.583541   \n",
       "4                                          0.522298                0.598783   \n",
       "\n",
       "   realized-sales-gross-margin  operating-profit-rate  \\\n",
       "0                     0.601457               0.998969   \n",
       "1                     0.610235               0.998946   \n",
       "2                     0.601364               0.998857   \n",
       "3                     0.583541               0.998700   \n",
       "4                     0.598783               0.998973   \n",
       "\n",
       "   pre-tax-net-interest-rate  after-tax-net-interest-rate  \\\n",
       "0                   0.796887                     0.808809   \n",
       "1                   0.797380                     0.809301   \n",
       "2                   0.796403                     0.808388   \n",
       "3                   0.796967                     0.808966   \n",
       "4                   0.797366                     0.809304   \n",
       "\n",
       "   non-industry-income-and-expenditure-revenue  ...  \\\n",
       "0                                     0.302646  ...   \n",
       "1                                     0.303556  ...   \n",
       "2                                     0.302035  ...   \n",
       "3                                     0.303350  ...   \n",
       "4                                     0.303475  ...   \n",
       "\n",
       "   net-income-to-total-assets  total-assets-to-gnp-price  no-credit-interval  \\\n",
       "0                    0.716845                   0.009219            0.622879   \n",
       "1                    0.795297                   0.008323            0.623652   \n",
       "2                    0.774670                   0.040003            0.623841   \n",
       "3                    0.739555                   0.003252            0.622929   \n",
       "4                    0.795016                   0.003878            0.623521   \n",
       "\n",
       "   gross-profit-to-sales  net-income-to-stockholder-s-equity  \\\n",
       "0               0.601453                            0.827890   \n",
       "1               0.610237                            0.839969   \n",
       "2               0.601449                            0.836774   \n",
       "3               0.583538                            0.834697   \n",
       "4               0.598782                            0.839973   \n",
       "\n",
       "   liability-to-equity  degree-of-financial-leverage-dfl  \\\n",
       "0             0.290202                          0.026601   \n",
       "1             0.283846                          0.264577   \n",
       "2             0.290189                          0.026555   \n",
       "3             0.281721                          0.026697   \n",
       "4             0.278514                          0.024752   \n",
       "\n",
       "   interest-coverage-ratio-interest-expense-to-ebit  net-income-flag  \\\n",
       "0                                          0.564050                1   \n",
       "1                                          0.570175                1   \n",
       "2                                          0.563706                1   \n",
       "3                                          0.564663                1   \n",
       "4                                          0.575617                1   \n",
       "\n",
       "   equity-to-liability  \n",
       "0             0.016469  \n",
       "1             0.020794  \n",
       "2             0.016474  \n",
       "3             0.023982  \n",
       "4             0.035490  \n",
       "\n",
       "[5 rows x 96 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from slugify import slugify\n",
    "\n",
    "columns = [slugify(name) for name in df.columns.values]\n",
    "df.columns = columns\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "interim-danger",
   "metadata": {},
   "source": [
    "Now I'm getting **target column** and the **rest of the columns** as features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "structural-probability",
   "metadata": {},
   "outputs": [],
   "source": [
    "target   = 'bankrupt'\n",
    "features = [name for name in df.columns.values if name != 'bankrupt']"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "informational-popularity",
   "metadata": {},
   "source": [
    "## RESAMPLING"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "joint-evidence",
   "metadata": {},
   "outputs": [],
   "source": [
    "from imblearn.over_sampling import SMOTE\n",
    "\n",
    "X = df[features]\n",
    "y = df[target]\n",
    "\n",
    "# RESAMPLING\n",
    "rnd = SMOTE()\n",
    "X, y = rnd.fit_resample(X, y)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "raising-administrator",
   "metadata": {},
   "source": [
    "## SCALE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "sized-translator",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "X_scaled = StandardScaler().fit_transform(X)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "elegant-halloween",
   "metadata": {},
   "source": [
    "## FEATURE SELECTION"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "becoming-omaha",
   "metadata": {},
   "source": [
    "Creating initial target/features selection. I tried ALL but even a trivial matrix plot was taking forever. That's why I decided to do a top 10 feature selection directly."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "adult-excellence",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/lib/python3.8/site-packages/sklearn/feature_selection/_univariate_selection.py:114: UserWarning: Features [93] are constant.\n",
      "  warnings.warn(\"Features %s are constant.\" % constant_features_idx,\n",
      "/opt/conda/lib/python3.8/site-packages/sklearn/feature_selection/_univariate_selection.py:116: RuntimeWarning: invalid value encountered in true_divide\n",
      "  f = msb / msw\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "['roa-c-before-interest-and-depreciation-before-interest',\n",
       " 'roa-a-before-interest-and-after-tax',\n",
       " 'tax-rate-a',\n",
       " 'net-value-per-share-c',\n",
       " 'operating-profit-per-share-yuan-y',\n",
       " 'total-debt-total-net-worth',\n",
       " 'debt-ratio',\n",
       " 'operating-profit-paid-in-capital',\n",
       " 'liability-assets-flag']"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.feature_selection import SelectKBest\n",
    "from sklearn.feature_selection import f_classif\n",
    "\n",
    "fs            = SelectKBest(f_classif, k=10)\n",
    "X_scaled      = fs.fit_transform(X_scaled, y)\n",
    "cols          = fs.get_support(indices=True)\n",
    "cols          = [name for name in df.columns[cols] if name != 'bankrupt']\n",
    "cols"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "broad-rings",
   "metadata": {},
   "source": [
    "Now creating the scatter matrix to see feature distributions"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "electoral-transparency",
   "metadata": {},
   "source": [
    "## CROSS VALIDATION"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "consistent-chinese",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import cross_val_score, train_test_split\n",
    "\n",
    "X = X_scaled\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "emerging-mailman",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "LogisticRegression\n",
      "==================\n",
      "{'C': 1} 0.8890244129840692\n",
      "---\n",
      "KNeighborsClassifier\n",
      "====================\n",
      "{'n_neighbors': 5} 0.9584614619612848\n",
      "---\n",
      "DecisionTreeClassifier\n",
      "======================\n",
      "{'max_depth': 4} 0.8996129724667619\n",
      "---\n"
     ]
    }
   ],
   "source": [
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.svm import LinearSVC\n",
    "from sklearn.neural_network import MLPClassifier\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "def print_search_result(name, search_result):\n",
    "    print(name)\n",
    "    print(\"=\" * len(name))\n",
    "    print(search_result.best_params_, search_result.best_score_)\n",
    "    print('---')\n",
    "    \n",
    "classifiers = [\n",
    "    {\n",
    "        'classifier': LogisticRegression(),\n",
    "        'params': { 'C': [1, 5, 10, 20, 40] }\n",
    "    },\n",
    "    {\n",
    "        'classifier': KNeighborsClassifier(),\n",
    "        'params': { 'n_neighbors': [5, 10, 15] }\n",
    "    },\n",
    "    {\n",
    "        'classifier': DecisionTreeClassifier(),\n",
    "        'params': { 'max_depth': [3, 4, 5, 6] }\n",
    "    }\n",
    "]\n",
    "\n",
    "for entry in classifiers:\n",
    "    classifier = entry['classifier']\n",
    "    classifier_name = type(classifier).__name__\n",
    "    \n",
    "    search_params   = entry['params']\n",
    "    search_result   = GridSearchCV(classifier, param_grid=search_params, scoring='recall')\\\n",
    "        .fit(X_train, y_train)\n",
    "    \n",
    "    print_search_result(classifier_name, search_result)\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "boxed-worry",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "\n",
    "log = LogisticRegression(C=1).fit(X_train, y_train)\n",
    "knn = KNeighborsClassifier(n_neighbors=5).fit(X_train, y_train)\n",
    "dtc = DecisionTreeClassifier(max_depth=6).fit(X_train, y_train)\n",
    "\n",
    "lst = [log, knn, dtc]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "champion-control",
   "metadata": {},
   "source": [
    "## Decision functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "moral-papua",
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_y_predict(clsf, samples):\n",
    "    dec_fun = getattr(clsf, 'decision_function', None)\n",
    "    if dec_fun:\n",
    "        return clsf.decision_function(samples)\n",
    "    else:\n",
    "        # only interested in positive score\n",
    "        return clsf.predict_proba(samples)[0:, 1:]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "oriental-sharing",
   "metadata": {},
   "source": [
    "## Precision/Recall curves"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "native-restoration",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x288 with 3 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import numpy as np\n",
    "from sklearn.metrics import precision_recall_curve\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# threshold I'm interested in\n",
    "THRESHOLD = 0.75\n",
    "\n",
    "plt.figure()\n",
    "_, ax = plt.subplots(1, 3, figsize=(15, 4))\n",
    "cols  = 0\n",
    "\n",
    "# plotting precision-recall charts\n",
    "for classifier in lst:\n",
    "    y_predict = get_y_predict(classifier, X_test)\n",
    "    precision, recall, thresholds = precision_recall_curve(y_test, y_predict)\n",
    "    ax[cols].title.set_text(str(type(classifier).__name__))\n",
    "    ax[cols].set(xlabel='Precision', ylabel='Recall')\n",
    "    ax[cols].step(precision, recall)\n",
    "    criteria = np.argmin(np.abs(thresholds - THRESHOLD))\n",
    "    ax[cols].plot(precision[criteria], recall[criteria], 'o', c='r')\n",
    "    ax[cols].grid(axis='both', linestyle='--', c='#cccccc')\n",
    "    cols+=1\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "printable-eight",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}