mda.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 265,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "import scipy\n",
    "\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 266,
   "metadata": {},
   "outputs": [],
   "source": [
    "from numpy.random import seed\n",
    "from numpy.random import randn\n",
    "from scipy.stats import anderson\n",
    "\n",
    "from scipy.stats import boxcox\n",
    "from scipy.special import inv_boxcox\n",
    "\n",
    "from numpy.random import seed\n",
    "from numpy.random import randn\n",
    "from scipy.stats import shapiro"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 267,
   "metadata": {},
   "outputs": [],
   "source": [
    "inspector_results = pd.read_csv(\"mda_dataset.csv\")\n",
    "l2_results = pd.read_csv(\"l2_analyzer_essay.csv\")\n",
    "taassc_results = pd.read_csv(\"result_taassc_essay.csv\")\n",
    "errors_data = pd.read_excel(\"dannye_novye.xlsx\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 268,
   "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>text_name</th>\n",
       "      <th>errors</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>exam2014/AAl_11_1.txt</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>exam2014/AAl_20_1.txt</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               text_name  errors\n",
       "0  exam2014/AAl_11_1.txt     1.0\n",
       "1  exam2014/AAl_20_1.txt     1.0"
      ]
     },
     "execution_count": 268,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "errors_data = errors_data.iloc[:-2]\n",
    "errors_data = errors_data[[\"text_name\", \"errors\"]]\n",
    "errors_data.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 269,
   "metadata": {},
   "outputs": [],
   "source": [
    "def filter_num_words(text):\n",
    "    if len(text.split()) >= 60:\n",
    "        return True\n",
    "    else:\n",
    "        return False\n",
    "    \n",
    "inspector_results = inspector_results[inspector_results[\"type\"] == 2]\n",
    "del inspector_results[\"type\"]\n",
    "filter_arr = [filter_num_words(val) for val in inspector_results[\"text\"].values]\n",
    "inspector_results = inspector_results[filter_arr]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 270,
   "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>av_depth</th>\n",
       "      <th>max_depth</th>\n",
       "      <th>min_depth</th>\n",
       "      <th>num_acl</th>\n",
       "      <th>num_rel_cl</th>\n",
       "      <th>num_advcl</th>\n",
       "      <th>num_sent</th>\n",
       "      <th>num_tok</th>\n",
       "      <th>av_tok_before_root</th>\n",
       "      <th>av_len_sent</th>\n",
       "      <th>...</th>\n",
       "      <th>punct_mistakes_pp</th>\n",
       "      <th>punct_mistakes_because</th>\n",
       "      <th>punct_mistakes_but</th>\n",
       "      <th>punct_mistakes_compare</th>\n",
       "      <th>million_mistake</th>\n",
       "      <th>side_mistake</th>\n",
       "      <th>name</th>\n",
       "      <th>text</th>\n",
       "      <th>class</th>\n",
       "      <th>sum_punct</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3.81</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>13</td>\n",
       "      <td>16</td>\n",
       "      <td>292</td>\n",
       "      <td>4.56</td>\n",
       "      <td>18.25</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>./data/exam/exam2017/DOv_2_2</td>\n",
       "      <td>More and more young people are stunding on the...</td>\n",
       "      <td>12</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4.33</td>\n",
       "      <td>7</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>6</td>\n",
       "      <td>12</td>\n",
       "      <td>221</td>\n",
       "      <td>6.67</td>\n",
       "      <td>18.42</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>./data/exam/exam2017/EGe_61_2</td>\n",
       "      <td>﻿Nowdays we can notice the tendency that large...</td>\n",
       "      <td>13</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>2 rows × 76 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   av_depth  max_depth  min_depth  num_acl  num_rel_cl  num_advcl  num_sent  \\\n",
       "2      3.81          6          0        5           0         13        16   \n",
       "3      4.33          7          2        3           3          6        12   \n",
       "\n",
       "   num_tok  av_tok_before_root  av_len_sent  ...  punct_mistakes_pp  \\\n",
       "2      292                4.56        18.25  ...                  0   \n",
       "3      221                6.67        18.42  ...                  0   \n",
       "\n",
       "   punct_mistakes_because  punct_mistakes_but  punct_mistakes_compare  \\\n",
       "2                       0                   1                       0   \n",
       "3                       2                   0                       0   \n",
       "\n",
       "   million_mistake  side_mistake                           name  \\\n",
       "2                0             0   ./data/exam/exam2017/DOv_2_2   \n",
       "3                0             0  ./data/exam/exam2017/EGe_61_2   \n",
       "\n",
       "                                                text  class  sum_punct  \n",
       "2  More and more young people are stunding on the...     12          1  \n",
       "3  ﻿Nowdays we can notice the tendency that large...     13          2  \n",
       "\n",
       "[2 rows x 76 columns]"
      ]
     },
     "execution_count": 270,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "inspector_results.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 271,
   "metadata": {},
   "outputs": [],
   "source": [
    "def fix_name(x):\n",
    "    if x[20] == \"_\":\n",
    "        x = x[:20] + x[22:]\n",
    "    return x[12:] + \".txt\"\n",
    "\n",
    "    \n",
    "inspector_results[\"text_name\"] = inspector_results[\"name\"].apply(fix_name)\n",
    "# inspector_results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 272,
   "metadata": {},
   "outputs": [],
   "source": [
    "full_data = inspector_results.merge(errors_data, on=\"text_name\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 273,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "\n",
    "for i, text in enumerate(full_data[\"text\"].values):\n",
    "\n",
    "    with open(f\"text/{i}.txt\", \"w\") as f:\n",
    "        f.write(text)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 274,
   "metadata": {},
   "outputs": [],
   "source": [
    "taassc_results[\"FILENAME\"] = taassc_results[\"filename\"]\n",
    "del taassc_results[\"filename\"]\n",
    "# taassc_results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 275,
   "metadata": {},
   "outputs": [],
   "source": [
    "def fix_name(x):\n",
    "    if x[20] == \"_\":\n",
    "        x = x[:20] + x[22:]\n",
    "    return x[12:] + \".txt\"\n",
    "    \n",
    "    \n",
    "inspector_results[\"text_name\"] = inspector_results[\"name\"].apply(fix_name)\n",
    "# inspector_results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 276,
   "metadata": {},
   "outputs": [],
   "source": [
    "inspector_results[\"FILENAME\"] = [str(i) + \".txt\" for i in list(range(len(inspector_results)))]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 277,
   "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>Unnamed: 0</th>\n",
       "      <th>FILENAME</th>\n",
       "      <th>W</th>\n",
       "      <th>S</th>\n",
       "      <th>VP</th>\n",
       "      <th>C</th>\n",
       "      <th>T</th>\n",
       "      <th>DC</th>\n",
       "      <th>CT</th>\n",
       "      <th>CP</th>\n",
       "      <th>...</th>\n",
       "      <th>VP/T</th>\n",
       "      <th>C/T</th>\n",
       "      <th>DC/C</th>\n",
       "      <th>DC/T</th>\n",
       "      <th>T/S</th>\n",
       "      <th>CT/T</th>\n",
       "      <th>CP/T</th>\n",
       "      <th>CP/C</th>\n",
       "      <th>CN/T</th>\n",
       "      <th>CN/C</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>936.txt</td>\n",
       "      <td>281</td>\n",
       "      <td>15</td>\n",
       "      <td>36</td>\n",
       "      <td>27</td>\n",
       "      <td>16</td>\n",
       "      <td>11</td>\n",
       "      <td>9</td>\n",
       "      <td>4</td>\n",
       "      <td>...</td>\n",
       "      <td>2.2500</td>\n",
       "      <td>1.6875</td>\n",
       "      <td>0.4074</td>\n",
       "      <td>0.6875</td>\n",
       "      <td>1.0667</td>\n",
       "      <td>0.5625</td>\n",
       "      <td>0.2500</td>\n",
       "      <td>0.1481</td>\n",
       "      <td>3.0625</td>\n",
       "      <td>1.8148</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>937.txt</td>\n",
       "      <td>345</td>\n",
       "      <td>17</td>\n",
       "      <td>48</td>\n",
       "      <td>37</td>\n",
       "      <td>19</td>\n",
       "      <td>17</td>\n",
       "      <td>14</td>\n",
       "      <td>5</td>\n",
       "      <td>...</td>\n",
       "      <td>2.5263</td>\n",
       "      <td>1.9474</td>\n",
       "      <td>0.4595</td>\n",
       "      <td>0.8947</td>\n",
       "      <td>1.1176</td>\n",
       "      <td>0.7368</td>\n",
       "      <td>0.2632</td>\n",
       "      <td>0.1351</td>\n",
       "      <td>3.2632</td>\n",
       "      <td>1.6757</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>2 rows × 25 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   Unnamed: 0 FILENAME    W   S  VP   C   T  DC  CT  CP  ...    VP/T     C/T  \\\n",
       "0           0  936.txt  281  15  36  27  16  11   9   4  ...  2.2500  1.6875   \n",
       "1           1  937.txt  345  17  48  37  19  17  14   5  ...  2.5263  1.9474   \n",
       "\n",
       "     DC/C    DC/T     T/S    CT/T    CP/T    CP/C    CN/T    CN/C  \n",
       "0  0.4074  0.6875  1.0667  0.5625  0.2500  0.1481  3.0625  1.8148  \n",
       "1  0.4595  0.8947  1.1176  0.7368  0.2632  0.1351  3.2632  1.6757  \n",
       "\n",
       "[2 rows x 25 columns]"
      ]
     },
     "execution_count": 277,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "l2_results.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 278,
   "metadata": {},
   "outputs": [],
   "source": [
    "full_data = taassc_results.merge(inspector_results.merge(l2_results, on=\"FILENAME\"), on=\"FILENAME\") \\\n",
    "            .merge(errors_data, on=\"text_name\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 279,
   "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>nwords</th>\n",
       "      <th>av_nominal_deps</th>\n",
       "      <th>av_nsubj_deps</th>\n",
       "      <th>av_nsubj_pass_deps</th>\n",
       "      <th>av_agents_deps</th>\n",
       "      <th>av_dobj_deps</th>\n",
       "      <th>av_pobj_deps</th>\n",
       "      <th>av_iobj_deps</th>\n",
       "      <th>av_ncomp_deps</th>\n",
       "      <th>av_nominal_deps_NN</th>\n",
       "      <th>...</th>\n",
       "      <th>C/T</th>\n",
       "      <th>DC/C</th>\n",
       "      <th>DC/T</th>\n",
       "      <th>T/S</th>\n",
       "      <th>CT/T</th>\n",
       "      <th>CP/T</th>\n",
       "      <th>CP/C</th>\n",
       "      <th>CN/T</th>\n",
       "      <th>CN/C</th>\n",
       "      <th>errors</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>256</td>\n",
       "      <td>0.737500</td>\n",
       "      <td>0.685714</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.941176</td>\n",
       "      <td>0.692308</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.783784</td>\n",
       "      <td>...</td>\n",
       "      <td>2.0000</td>\n",
       "      <td>0.5</td>\n",
       "      <td>1.0000</td>\n",
       "      <td>1.0000</td>\n",
       "      <td>0.8125</td>\n",
       "      <td>0.3750</td>\n",
       "      <td>0.1875</td>\n",
       "      <td>2.3125</td>\n",
       "      <td>1.1562</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>401</td>\n",
       "      <td>0.889764</td>\n",
       "      <td>0.770833</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.212121</td>\n",
       "      <td>0.731707</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.666667</td>\n",
       "      <td>1.018349</td>\n",
       "      <td>...</td>\n",
       "      <td>1.5517</td>\n",
       "      <td>0.4</td>\n",
       "      <td>0.6207</td>\n",
       "      <td>1.0741</td>\n",
       "      <td>0.5517</td>\n",
       "      <td>0.4828</td>\n",
       "      <td>0.3111</td>\n",
       "      <td>1.9655</td>\n",
       "      <td>1.2667</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>2 rows × 268 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   nwords  av_nominal_deps  av_nsubj_deps  av_nsubj_pass_deps  av_agents_deps  \\\n",
       "0     256         0.737500       0.685714                 0.5             0.0   \n",
       "1     401         0.889764       0.770833                 0.5             0.0   \n",
       "\n",
       "   av_dobj_deps  av_pobj_deps  av_iobj_deps  av_ncomp_deps  \\\n",
       "0      0.941176      0.692308           0.0       0.000000   \n",
       "1      1.212121      0.731707           0.0       1.666667   \n",
       "\n",
       "   av_nominal_deps_NN  ...     C/T  DC/C    DC/T     T/S    CT/T    CP/T  \\\n",
       "0            0.783784  ...  2.0000   0.5  1.0000  1.0000  0.8125  0.3750   \n",
       "1            1.018349  ...  1.5517   0.4  0.6207  1.0741  0.5517  0.4828   \n",
       "\n",
       "     CP/C    CN/T    CN/C  errors  \n",
       "0  0.1875  2.3125  1.1562     1.0  \n",
       "1  0.3111  1.9655  1.2667     4.0  \n",
       "\n",
       "[2 rows x 268 columns]"
      ]
     },
     "execution_count": 279,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "full_data.head(2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Выделим только необходимые метрики из каждого инструмента:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 280,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "226"
      ]
     },
     "execution_count": 280,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "s = \"av_depth max_depth min_depth num_acl num_advcl num_sent num_tok av_tok_before_root av_len_sent num_cl num_tu num_compl_tu num_coord num_adj_noun num_part_noun num_noun_inf pos_sim_nei lemma_sim_nei pos_sim_all lemma_sim_all\"\n",
    "cols = [st for st in s.split()] + [\"class\"]\n",
    "\n",
    "l2_must_have = ['MLS',\n",
    "       'MLT', 'MLC', 'C/S', 'VP/T', 'C/T', 'DC/C', 'DC/T', 'T/S', 'CT/T',\n",
    "       'CP/T', 'CP/C', 'CN/T', 'CN/C']\n",
    "\n",
    "taassc_must_have = \"\"\"acad_av_lemma_freq\n",
    "acad_av_construction_freq\n",
    "acad_av_lemma_construction_freq\n",
    "acad_av_approx_collexeme\n",
    "acad_av_faith_verb_cue\n",
    "acad_av_faith_const_cue\n",
    "acad_av_delta_p_verb_cue\n",
    "acad_av_delta_p_const_cue\n",
    "acad_av_lemma_freq_log\n",
    "acad_av_construction_freq_log\n",
    "acad_av_lemma_construction_freq_log\n",
    "acad_av_lemma_freq_type\n",
    "acad_av_construction_freq_type\n",
    "acad_av_lemma_construction_freq_type\n",
    "acad_av_approx_collexeme_type\n",
    "acad_av_faith_verb_cue_type\n",
    "acad_av_faith_const_cue_type\n",
    "acad_av_delta_p_verb_cue_type\n",
    "acad_av_delta_p_const_cue_type\n",
    "acad_collexeme_ratio\n",
    "acad_collexeme_ratio_type\n",
    "acad_lemma_ttr\n",
    "acad_construction_ttr\n",
    "acad_lemma_construction_ttr\n",
    "acad_lemma_attested\n",
    "acad_construction_attested\n",
    "acad_lemma_construction_attested\n",
    "news_av_lemma_freq\n",
    "news_av_construction_freq\n",
    "news_av_lemma_construction_freq\n",
    "news_av_approx_collexeme\n",
    "news_av_faith_verb_cue\n",
    "news_av_faith_const_cue\n",
    "news_av_delta_p_verb_cue\n",
    "news_av_delta_p_const_cue\n",
    "news_av_lemma_freq_log\n",
    "news_av_construction_freq_log\n",
    "news_av_lemma_construction_freq_log\n",
    "news_av_lemma_freq_type\n",
    "news_av_construction_freq_type\n",
    "news_av_lemma_construction_freq_type\n",
    "news_av_approx_collexeme_type\n",
    "news_av_faith_verb_cue_type\n",
    "news_av_faith_const_cue_type\n",
    "news_av_delta_p_verb_cue_type\n",
    "news_av_delta_p_const_cue_type\n",
    "news_collexeme_ratio\n",
    "news_collexeme_ratio_type\n",
    "news_lemma_ttr\n",
    "news_construction_ttr\n",
    "news_lemma_construction_ttr\n",
    "news_lemma_attested\n",
    "news_construction_attested\n",
    "news_lemma_construction_attested\n",
    "mag_av_lemma_freq\n",
    "mag_av_construction_freq\n",
    "mag_av_lemma_construction_freq\n",
    "mag_av_approx_collexeme\n",
    "mag_av_faith_verb_cue\n",
    "mag_av_faith_const_cue\n",
    "mag_av_delta_p_verb_cue\n",
    "mag_av_delta_p_const_cue\n",
    "mag_av_lemma_freq_log\n",
    "mag_av_construction_freq_log\n",
    "mag_av_lemma_construction_freq_log\n",
    "mag_av_lemma_freq_type\n",
    "mag_av_construction_freq_type\n",
    "mag_av_lemma_construction_freq_type\n",
    "mag_av_approx_collexeme_type\n",
    "mag_av_faith_verb_cue_type\n",
    "mag_av_faith_const_cue_type\n",
    "mag_av_delta_p_verb_cue_type\n",
    "mag_av_delta_p_const_cue_type\n",
    "mag_collexeme_ratio\n",
    "mag_collexeme_ratio_type\n",
    "mag_lemma_ttr\n",
    "mag_construction_ttr\n",
    "mag_lemma_construction_ttr\n",
    "mag_lemma_attested\n",
    "mag_construction_attested\n",
    "mag_lemma_construction_attested\n",
    "fic_av_lemma_freq\n",
    "fic_av_construction_freq\n",
    "fic_av_lemma_construction_freq\n",
    "fic_av_approx_collexeme\n",
    "fic_av_faith_verb_cue\n",
    "fic_av_faith_const_cue\n",
    "fic_av_delta_p_verb_cue\n",
    "fic_av_delta_p_const_cue\n",
    "fic_av_lemma_freq_log\n",
    "fic_av_construction_freq_log\n",
    "fic_av_lemma_construction_freq_log\n",
    "fic_av_lemma_freq_type\n",
    "fic_av_construction_freq_type\n",
    "fic_av_lemma_construction_freq_type\n",
    "fic_av_approx_collexeme_type\n",
    "fic_av_faith_verb_cue_type\n",
    "fic_av_faith_const_cue_type\n",
    "fic_av_delta_p_verb_cue_type\n",
    "fic_av_delta_p_const_cue_type\n",
    "fic_collexeme_ratio\n",
    "fic_collexeme_ratio_type\n",
    "fic_lemma_ttr\n",
    "fic_construction_ttr\n",
    "fic_lemma_construction_ttr\n",
    "fic_lemma_attested\n",
    "fic_construction_attested\n",
    "fic_lemma_construction_attested\n",
    "all_av_lemma_freq\n",
    "all_av_construction_freq\n",
    "all_av_lemma_construction_freq\n",
    "all_av_approx_collexeme\n",
    "all_av_faith_verb_cue\n",
    "all_av_faith_const_cue\n",
    "all_av_delta_p_verb_cue\n",
    "all_av_delta_p_const_cue\n",
    "all_av_lemma_freq_log\n",
    "all_av_construction_freq_log\n",
    "all_av_lemma_construction_freq_log\n",
    "all_av_lemma_freq_type\n",
    "all_av_construction_freq_type\n",
    "all_av_lemma_construction_freq_type\n",
    "all_av_approx_collexeme_type\n",
    "all_av_faith_verb_cue_type\n",
    "all_av_faith_const_cue_type\n",
    "all_av_delta_p_verb_cue_type\n",
    "all_av_delta_p_const_cue_type\n",
    "all_collexeme_ratio\n",
    "all_collexeme_ratio_type\n",
    "all_lemma_ttr\n",
    "all_construction_ttr\n",
    "all_lemma_construction_ttr\n",
    "all_lemma_attested\n",
    "all_construction_attested\n",
    "all_lemma_construction_attested\n",
    "all_av_lemma_freq_stdev\n",
    "all_av_construction_freq_stdev\n",
    "all_av_lemma_construction_freq_stdev\n",
    "all_av_approx_collexeme_stdev\n",
    "all_av_faith_verb_cue_stdev\n",
    "all_av_faith_const_cue_stdev\n",
    "all_av_delta_p_verb_cue_stdev\n",
    "all_av_delta_p_const_cue_stdev\n",
    "acad_av_lemma_freq_stdev\n",
    "acad_av_construction_freq_stdev\n",
    "acad_av_lemma_construction_freq_stdev\n",
    "acad_av_approx_collexeme_stdev\n",
    "acad_av_faith_verb_cue_stdev\n",
    "acad_av_faith_const_cue_stdev\n",
    "acad_av_delta_p_verb_cue_stdev\n",
    "acad_av_delta_p_const_cue_stdev\n",
    "news_av_lemma_freq_stdev\n",
    "news_av_construction_freq_stdev\n",
    "news_av_lemma_construction_freq_stdev\n",
    "news_av_approx_collexeme_stdev\n",
    "news_av_faith_verb_cue_stdev\n",
    "news_av_faith_const_cue_stdev\n",
    "news_av_delta_p_verb_cue_stdev\n",
    "news_av_delta_p_const_cue_stdev\n",
    "mag_av_lemma_freq_stdev\n",
    "mag_av_construction_freq_stdev\n",
    "mag_av_lemma_construction_freq_stdev\n",
    "mag_av_approx_collexeme_stdev\n",
    "mag_av_faith_verb_cue_stdev\n",
    "mag_av_faith_const_cue_stdev\n",
    "mag_av_delta_p_verb_cue_stdev\n",
    "mag_av_delta_p_const_cue_stdev\n",
    "fic_av_lemma_freq_stdev\n",
    "fic_av_construction_freq_stdev\n",
    "fic_av_lemma_construction_freq_stdev\n",
    "fic_av_approx_collexeme_stdev\n",
    "fic_av_faith_verb_cue_stdev\n",
    "fic_av_faith_const_cue_stdev\n",
    "fic_av_delta_p_verb_cue_stdev\n",
    "fic_av_delta_p_const_cue_stdev\n",
    "all_av_lemma_freq_log_stdev\n",
    "all_av_construction_freq_log_stdev\n",
    "all_av_lemma_construction_freq_log_stdev\n",
    "acad_av_lemma_freq_log_stdev\n",
    "acad_av_construction_freq_log_stdev\n",
    "acad_av_lemma_construction_freq_log_stdev\n",
    "news_av_lemma_freq_log_stdev\n",
    "news_av_construction_freq_log_stdev\n",
    "news_av_lemma_construction_freq_log_stdev\n",
    "mag_av_lemma_freq_log_stdev\n",
    "mag_av_construction_freq_log_stdev\n",
    "mag_av_lemma_construction_freq_log_stdev\n",
    "fic_av_lemma_freq_log_stdev\n",
    "fic_av_construction_freq_log_stdev\n",
    "fic_av_lemma_construction_freq_log_stdev\"\"\"\n",
    "\n",
    "taassc_must_have = taassc_must_have.split(\"\\n\")\n",
    "\n",
    "must_have_cols = cols + l2_must_have + list(taassc_must_have) + [\"errors\"]\n",
    "len(must_have_cols)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 281,
   "metadata": {},
   "outputs": [],
   "source": [
    "for col_name in full_data.columns:\n",
    "    if col_name not in must_have_cols:\n",
    "        del full_data[col_name]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Посмотрим, сколько эссе содержится в каждом классе"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 282,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{3, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16}\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[(3, 1),\n",
       " (6, 2),\n",
       " (7, 1),\n",
       " (8, 4),\n",
       " (9, 11),\n",
       " (10, 34),\n",
       " (11, 89),\n",
       " (12, 222),\n",
       " (13, 106),\n",
       " (14, 43),\n",
       " (15, 5),\n",
       " (16, 1)]"
      ]
     },
     "execution_count": 282,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(set(full_data[\"class\"].values))\n",
    "[(i, len(full_data[full_data[\"class\"] == i])) for i in \n",
    "     set(full_data[\"class\"].values)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 283,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{1.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, 12.0, 13.0, 14.0, 16.0}\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[(1.0, 290),\n",
       " (4.0, 86),\n",
       " (5.0, 45),\n",
       " (6.0, 35),\n",
       " (7.0, 23),\n",
       " (8.0, 15),\n",
       " (9.0, 9),\n",
       " (10.0, 4),\n",
       " (11.0, 4),\n",
       " (12.0, 2),\n",
       " (13.0, 1),\n",
       " (14.0, 3),\n",
       " (16.0, 2)]"
      ]
     },
     "execution_count": 283,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(set(full_data[\"errors\"].values))\n",
    "[(i, len(full_data[full_data[\"errors\"] == i])) for i in \n",
    "     set(full_data[\"errors\"].values)]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Отбросим группы, в которых слишком мало данных"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 284,
   "metadata": {},
   "outputs": [],
   "source": [
    "full_data_error = full_data.copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 285,
   "metadata": {},
   "outputs": [],
   "source": [
    "full_data_error = full_data_error[full_data_error[\"errors\"] <= 9]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 286,
   "metadata": {},
   "outputs": [],
   "source": [
    "full_data = full_data[full_data[\"class\"] != 3]\n",
    "full_data = full_data[full_data[\"class\"] != 4]\n",
    "full_data = full_data[full_data[\"class\"] != 5]\n",
    "full_data = full_data[full_data[\"class\"] != 6]\n",
    "full_data = full_data[full_data[\"class\"] != 7]\n",
    "full_data = full_data[full_data[\"class\"] != 8]\n",
    "full_data = full_data[full_data[\"class\"] != 15]\n",
    "full_data = full_data[full_data[\"class\"] != 16]\n",
    "full_data = full_data[full_data[\"class\"] != 17]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 287,
   "metadata": {},
   "outputs": [],
   "source": [
    "full_data_norm = full_data.copy()\n",
    "full_data_corr = full_data.copy()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Применим функции преобразования и посмотрим, распределения каких метрик можно привести к нормальному:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 288,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "av_depth yeojohnson 0.104\n",
      "av_tok_before_root yeojohnson 0.434\n",
      "av_len_sent yeojohnson 0.508\n",
      "num_adj_noun yeojohnson 0.061\n",
      "pos_sim_nei yeojohnson 0.442\n",
      "lemma_sim_nei yeojohnson 0.227\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/koteuka/anaconda3/lib/python3.7/site-packages/scipy/stats/morestats.py:1673: UserWarning: Input data for shapiro has range zero. The results may not be accurate.\n",
      "  warnings.warn(\"Input data for shapiro has range zero. The results \"\n",
      "/home/koteuka/anaconda3/lib/python3.7/site-packages/ipykernel_launcher.py:14: RuntimeWarning: divide by zero encountered in log\n",
      "  \n",
      "/home/koteuka/anaconda3/lib/python3.7/site-packages/ipykernel_launcher.py:14: RuntimeWarning: divide by zero encountered in reciprocal\n",
      "  \n",
      "/home/koteuka/anaconda3/lib/python3.7/site-packages/ipykernel_launcher.py:14: RuntimeWarning: invalid value encountered in reciprocal\n",
      "  \n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MLS log 0.134\n",
      "MLT 1/x 0.099\n",
      "MLC boxcox 0.866\n",
      "C/S yeojohnson 0.09\n",
      "VP/T yeojohnson 0.881\n",
      "C/T yeojohnson 0.339\n",
      "DC/C yeojohnson 0.418\n",
      "DC/T yeojohnson 0.202\n",
      "CT/T yeojohnson 0.452\n",
      "CP/T yeojohnson 0.517\n",
      "CP/C yeojohnson 0.662\n",
      "CN/T yeojohnson 0.862\n",
      "CN/C yeojohnson 0.711\n"
     ]
    }
   ],
   "source": [
    "from scipy.stats import yeojohnson\n",
    "\n",
    "\n",
    "transformations = [\"orig\", \"log\", \"sqrt\", \"1/x\", \"yeojohnson\", \"boxcox\"]\n",
    "count = 0\n",
    "normal_cols = [\"class\"]\n",
    "\n",
    "\n",
    "def get_p_value(func, col, double_ret=False):\n",
    "    try:\n",
    "        if double_ret:\n",
    "            transform, _ = func(col)\n",
    "        else:\n",
    "            transform = func(col)\n",
    "        if len(transform) != len(col):\n",
    "            return 0.0\n",
    "        stat, p = shapiro(transform)\n",
    "        \n",
    "#         print(p)\n",
    "    except:\n",
    "        return 0.0\n",
    "#     print(p)\n",
    "    if p > 0.9:\n",
    "        p = 0.0\n",
    "#     print(\"p-norm:\", p)\n",
    "    return p\n",
    "        \n",
    "\n",
    "\n",
    "for col_name in full_data_norm.columns:\n",
    "    col = full_data_norm[col_name].values\n",
    "    if type(col[0]) != str:\n",
    "        shapiro_p_val = []\n",
    "        \n",
    "        stat, p = shapiro(col)\n",
    "        if p == 1.0:\n",
    "        \n",
    "            p = 0.0\n",
    "        shapiro_p_val.append(p)\n",
    "\n",
    "        shapiro_p_val.append(get_p_value(np.log, col))\n",
    "\n",
    "        shapiro_p_val.append(get_p_value(np.sqrt, col))\n",
    "\n",
    "        shapiro_p_val.append(get_p_value(np.reciprocal, col))\n",
    "\n",
    "        shapiro_p_val.append(get_p_value(yeojohnson, col, True))\n",
    "\n",
    "        shapiro_p_val.append(get_p_value(boxcox, col, True))\n",
    "\n",
    "#         print(shapiro_p_val)\n",
    "        if np.max(shapiro_p_val) > 0.05:\n",
    "#             print(\"p\", np.max(shapiro_p_val))\n",
    "#             print(shapiro_p_val)\n",
    "            normal_cols.append(col_name)\n",
    "            argmax = np.argmax(shapiro_p_val)\n",
    "            print(col_name, transformations[argmax], round(np.max(shapiro_p_val), 3))\n",
    "            if transformations[argmax] == \"orig\":\n",
    "                full_data_norm[col_name] = col\n",
    "            elif transformations[argmax] == \"log\":\n",
    "                full_data_norm[col_name] = np.log(col)\n",
    "            elif transformations[argmax] == \"sqrt\":\n",
    "                full_data_norm[col_name] = np.sqrt(col)\n",
    "            elif transformations[argmax] == \"1/x\":\n",
    "                full_data_norm[col_name] = np.reciprocal(col)\n",
    "            elif transformations[argmax] == \"yeojohnson\":\n",
    "#                 print(len(full_data_norm[col_name].values), len(col), len(yeojohnson(col)))\n",
    "                full_data_norm[col_name] = yeojohnson(col)[0]\n",
    "            elif transformations[argmax] == \"boxcox\":\n",
    "                full_data_norm[col_name] = boxcox(col)[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 289,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['class', 'av_depth', 'av_tok_before_root', 'av_len_sent', 'num_adj_noun', 'pos_sim_nei', 'lemma_sim_nei', 'MLS', 'MLT', 'MLC', 'C/S', 'VP/T', 'C/T', 'DC/C', 'DC/T', 'CT/T', 'CP/T', 'CP/C', 'CN/T', 'CN/C']\n"
     ]
    }
   ],
   "source": [
    "print(normal_cols)\n",
    "\n",
    "for col_name in full_data_norm.columns:\n",
    "    if col_name not in normal_cols:\n",
    "        del full_data_norm[col_name]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Применим тест Левена и отберём подходящие значения"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 290,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2.341 \t 0.0406 \t av_depth\n",
      "0.29 \t 0.9188 \t av_tok_before_root\n",
      "2.01 \t 0.0758 \t av_len_sent\n",
      "1.153 \t 0.3315 \t num_adj_noun\n",
      "2.528 \t 0.0283 \t pos_sim_nei\n",
      "2.763 \t 0.0179 \t lemma_sim_nei\n",
      "nan \t nan \t class\n",
      "0.739 \t 0.5942 \t MLS\n",
      "1.966 \t 0.0822 \t MLT\n",
      "0.904 \t 0.4779 \t MLC\n",
      "0.723 \t 0.6066 \t C/S\n",
      "0.494 \t 0.7807 \t VP/T\n",
      "0.432 \t 0.8264 \t C/T\n",
      "0.911 \t 0.4734 \t DC/C\n",
      "0.675 \t 0.6425 \t DC/T\n",
      "1.245 \t 0.2869 \t CT/T\n",
      "0.798 \t 0.5515 \t CP/T\n",
      "0.359 \t 0.8763 \t CP/C\n",
      "2.383 \t 0.0375 \t CN/T\n",
      "1.708 \t 0.1312 \t CN/C\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/koteuka/anaconda3/lib/python3.7/site-packages/scipy/stats/morestats.py:2369: RuntimeWarning: invalid value encountered in double_scalars\n",
      "  W = numer / denom\n"
     ]
    }
   ],
   "source": [
    "from scipy.stats import bartlett\n",
    "from scipy.stats import levene\n",
    "\n",
    "\n",
    "class_val = set(full_data_norm[\"class\"].values)\n",
    "normal_cols = [\"class\"]\n",
    "\n",
    "for col_name in full_data_norm.columns:\n",
    "    col = full_data_norm[col_name].values\n",
    "    if type(col[0]) != str:\n",
    "        arr = []\n",
    "        for class_v in class_val:\n",
    "            arr.append(full_data_norm[full_data_norm[\"class\"] == class_v][col_name])\n",
    "#         print([len(i) for i in arr])\n",
    "        # Bartlett's test in Python with SciPy:\n",
    "#         stat, p = bartlett(ctrl, trt1, trt2)\n",
    "        stat, p = levene(*arr)\n",
    "        # Get the results:\n",
    "        print(round(stat, 3), \"\\t\", round(p, 4), \"\\t\", col_name)\n",
    "        if p > 0.05:\n",
    "            normal_cols.append(col_name)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 291,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['class', 'av_tok_before_root', 'av_len_sent', 'num_adj_noun', 'MLS', 'MLT', 'MLC', 'C/S', 'VP/T', 'C/T', 'DC/C', 'DC/T', 'CT/T', 'CP/T', 'CP/C', 'CN/C']\n"
     ]
    }
   ],
   "source": [
    "print(normal_cols)\n",
    "\n",
    "for col_name in full_data_norm.columns:\n",
    "    if col_name not in normal_cols:\n",
    "        del full_data_norm[col_name]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Удалим мультиколлинеарные переменные, то есть для тех независимых переменных, у которых абсолютное значение коллинеарности по Спирмену больше, чем 0.7 возьмём ту метрику, которая показывает максимальный коэффициент корреляции с зависимой переменной"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Сначала удалим коррелирующие переменные для зависимого столбца 'class':"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 308,
   "metadata": {},
   "outputs": [],
   "source": [
    "filtered_data = full_data_corr.copy()\n",
    "del filtered_data[\"errors\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 309,
   "metadata": {},
   "outputs": [],
   "source": [
    "col_to_del = set()\n",
    "# class_col_index = [i for i, j in enumerate(filtered_data.columns) if j == \"class\"][0]\n",
    "# corr_mat = filtered_data.corr(\"spearman\").values\n",
    "\n",
    "break_flag = True\n",
    "prev_len_val = 0\n",
    "\n",
    "while True:\n",
    "    class_col_index = [i for i, j in enumerate(filtered_data.columns) if j == \"class\"][0]\n",
    "    corr_mat = filtered_data.corr(\"spearman\").values\n",
    "    col_to_del = set()\n",
    "    \n",
    "    for i, col_name in enumerate(filtered_data.columns):\n",
    "        buff_arr = []\n",
    "        for j, col_name_2 in enumerate(filtered_data.columns):\n",
    "            if i == j:\n",
    "                continue\n",
    "            if corr_mat[i][j] > 0.7:\n",
    "                buff_arr.append((j, col_name_2))\n",
    "        if len(buff_arr) > 0:\n",
    "            buff_arr.append((i, col_name))\n",
    "            break\n",
    "    \n",
    "    if len(buff_arr) == 0:\n",
    "        break\n",
    "    max_val = 0.0\n",
    "    max_col = None\n",
    "    for j, col in buff_arr:\n",
    "        if corr_mat[class_col_index][j] > max_val:\n",
    "            max_val = corr_mat[class_col_index][j]\n",
    "            max_col = col\n",
    "    col_to_del = col_to_del.union(set([i for _, i in buff_arr]))\n",
    "#     print(col_to_del)\n",
    "    col_to_del -= set([max_col])\n",
    "    \n",
    "    for col in col_to_del:\n",
    "        del filtered_data[col]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 310,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['max_depth', 'min_depth', 'num_acl', 'num_advcl', 'num_sent', 'num_tok',\n",
       "       'av_tok_before_root', 'num_tu', 'num_coord', 'num_adj_noun',\n",
       "       'num_part_noun', 'num_noun_inf', 'class', 'MLT', 'MLC', 'C/T', 'T/S',\n",
       "       'CP/T'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 310,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "filtered_data.columns"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Рассмотрим матрицу корреляции Спирмена, чтобы убедиться в отсутствии мультиколлинеарности между независимыми переменными"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 311,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1512x1224 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "f = plt.figure(figsize=(21, 17))\n",
    "plt.matshow(filtered_data.corr(\"spearman\"), fignum=f.number)\n",
    "plt.xticks(range(filtered_data.select_dtypes(['number']).shape[1]),\n",
    "           filtered_data.select_dtypes(['number']).columns, fontsize=14, rotation=90)\n",
    "plt.yticks(range(filtered_data.select_dtypes(['number']).shape[1]),\n",
    "           filtered_data.select_dtypes(['number']).columns, fontsize=14)\n",
    "cb = plt.colorbar()\n",
    "cb.ax.tick_params(labelsize=14)\n",
    "plt.title(' ', fontsize=16);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Отсортируем значения коэффициента корреляции с зависимой переменной для оценки:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 312,
   "metadata": {},
   "outputs": [],
   "source": [
    "class_col_index = [i for i, j in enumerate(filtered_data.columns) if j == \"class\"][0]\n",
    "\n",
    "\n",
    "corr_mat = filtered_data.corr(\"spearman\").values[class_col_index, :]\n",
    "corr_mat_kend = filtered_data.corr(\"kendall\").values[class_col_index, :]\n",
    "res_dict = {}\n",
    "for i, colname in enumerate(filtered_data.corr(\"spearman\").columns):\n",
    "    res_dict[colname] = (round(corr_mat[i], 3), round(corr_mat_kend[i], 3))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 313,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\t (Spearman, Kendal)\n",
      "\n",
      "class :   (1.0, 1.0)\n",
      "num_tok :   (0.439, 0.342)\n",
      "num_acl :   (0.295, 0.235)\n",
      "num_advcl :   (0.275, 0.215)\n",
      "num_adj_noun :   (0.257, 0.199)\n",
      "num_part_noun :   (0.245, 0.2)\n",
      "max_depth :   (0.235, 0.19)\n",
      "MLT :   (0.234, 0.175)\n",
      "num_sent :   (0.19, 0.151)\n",
      "num_noun_inf :   (0.177, 0.144)\n",
      "MLC :   (0.177, 0.133)\n",
      "num_tu :   (0.164, 0.126)\n",
      "av_tok_before_root :   (0.137, 0.103)\n",
      "num_coord :   (0.125, 0.098)\n",
      "min_depth :   (0.124, 0.104)\n",
      "CP/T :   (0.117, 0.087)\n",
      "C/T :   (0.108, 0.082)\n",
      "T/S :   (0.035, 0.027)\n"
     ]
    }
   ],
   "source": [
    "print(\"\\t (Spearman, Kendal)\\n\")\n",
    "\n",
    "for col, val in sorted(res_dict.items(), key=lambda x: -np.abs(x[1][0])):\n",
    "    print(col,\":  \", val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 314,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Colname:  Spearman p-value:   Kendall p-value:\n",
      "max_depth        0.0        0.0\n",
      "min_depth        0.0052        0.0055\n",
      "num_acl        0.0        0.0\n",
      "num_advcl        0.0        0.0\n",
      "num_sent        0.0        0.0\n",
      "num_tok        0.0        0.0\n",
      "av_tok_before_root        0.0021        0.002\n",
      "num_tu        0.0002        0.0002\n",
      "num_coord        0.0049        0.0044\n",
      "num_adj_noun        0.0        0.0\n",
      "num_part_noun        0.0        0.0\n",
      "num_noun_inf        0.0001        0.0001\n",
      "class        0.0        0.0\n",
      "MLT        0.0        0.0\n",
      "MLC        0.0001        0.0001\n",
      "C/T        0.0148        0.0144\n",
      "T/S        0.4317        0.424\n",
      "CP/T        0.0084        0.0091\n"
     ]
    }
   ],
   "source": [
    "from scipy import stats\n",
    "\n",
    "print(\"Colname:  Spearman p-value:   Kendall p-value:\")\n",
    "for col in filtered_data.columns:\n",
    "    val = filtered_data[col].values\n",
    "    print(col, \"      \", round(stats.spearmanr(val, filtered_data[\"class\"].values)[1], 4), \"      \", \n",
    "    round(stats.kendalltau(val, filtered_data[\"class\"].values)[1], 4))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Рассмотрим график зависимости среднего значения первых в рейтинге метрик от оценки:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 299,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'num_adj_noun')"
      ]
     },
     "execution_count": 299,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x576 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from matplotlib.pyplot import figure\n",
    "\n",
    "\n",
    "fig, axs = plt.subplots(2, 2, figsize=(14, 8))\n",
    "\n",
    "# fig.figure(figsize=(8, 6), dpi=80)\n",
    "fig.suptitle('Top metric mean value')\n",
    "\n",
    "axs[0][0].plot(list(set(filtered_data[\"class\"].values)),\n",
    "            filtered_data.groupby('class')[\"num_tok\"].mean().values)\n",
    "axs[0, 0].set_title(\"num_tok\")\n",
    "\n",
    "axs[0][1].plot(list(set(filtered_data[\"class\"].values)),\n",
    "            filtered_data.groupby('class')[\"num_acl\"].mean().values)\n",
    "axs[0, 1].set_title(\"num_tok\")\n",
    "\n",
    "axs[1][0].plot(list(set(filtered_data[\"class\"].values)),\n",
    "            filtered_data.groupby('class')[\"num_advcl\"].mean().values)\n",
    "axs[1, 0].set_title(\"num_advcl\")\n",
    "\n",
    "axs[1][1].plot(list(set(filtered_data[\"class\"].values)),\n",
    "            filtered_data.groupby('class')[\"num_adj_noun\"].mean().values)\n",
    "axs[1, 1].set_title(\"num_adj_noun\")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Сделаем те же самые шаги для выявления корреляции с количеством синтаксических ошибок:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 300,
   "metadata": {},
   "outputs": [],
   "source": [
    "filtered_data = full_data_error.copy()\n",
    "del filtered_data[\"class\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 301,
   "metadata": {},
   "outputs": [],
   "source": [
    "while True:\n",
    "    class_col_index = [i for i, j in enumerate(filtered_data.columns) if j == \"errors\"][0]\n",
    "    corr_mat = filtered_data.corr(\"spearman\").values\n",
    "    col_to_del = set()\n",
    "    \n",
    "    for i, col_name in enumerate(filtered_data.columns):\n",
    "        buff_arr = []\n",
    "        for j, col_name_2 in enumerate(filtered_data.columns):\n",
    "            if i == j:\n",
    "                continue\n",
    "            if corr_mat[i][j] > 0.7:\n",
    "                buff_arr.append((j, col_name_2))\n",
    "        if len(buff_arr) > 0:\n",
    "            buff_arr.append((i, col_name))\n",
    "            break\n",
    "    \n",
    "    if len(buff_arr) == 0:\n",
    "        break\n",
    "    max_val = 0.0\n",
    "    max_col = None\n",
    "    for j, col in buff_arr:\n",
    "        if corr_mat[class_col_index][j] > max_val:\n",
    "            max_val = corr_mat[class_col_index][j]\n",
    "            max_col = col\n",
    "    col_to_del = col_to_del.union(set([i for _, i in buff_arr]))\n",
    "#     print(col_to_del)\n",
    "    col_to_del -= set([max_col])\n",
    "    \n",
    "    for col in col_to_del:\n",
    "        del filtered_data[col]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Посмотрим матрицу корреляции"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 302,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1512x1224 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "f = plt.figure(figsize=(21, 17))\n",
    "plt.matshow(filtered_data.corr(), fignum=f.number)\n",
    "plt.xticks(range(filtered_data.select_dtypes(['number']).shape[1]),\n",
    "           filtered_data.select_dtypes(['number']).columns, fontsize=14, rotation=90)\n",
    "plt.yticks(range(filtered_data.select_dtypes(['number']).shape[1]),\n",
    "           filtered_data.select_dtypes(['number']).columns, fontsize=14)\n",
    "cb = plt.colorbar()\n",
    "cb.ax.tick_params(labelsize=14)\n",
    "plt.title(' ', fontsize=16);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Отсортируем значения коэффициента корреляции с зависимой переменной для количества синтаксических ошибок:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 303,
   "metadata": {},
   "outputs": [],
   "source": [
    "class_col_index = [i for i, j in enumerate(filtered_data.columns) if j == \"errors\"][0]\n",
    "\n",
    "\n",
    "corr_mat = filtered_data.corr(\"spearman\").values[class_col_index, :]\n",
    "corr_mat_kend = filtered_data.corr(\"kendall\").values[class_col_index, :]\n",
    "res_dict = {}\n",
    "for i, colname in enumerate(filtered_data.corr(\"spearman\").columns):\n",
    "    res_dict[colname] = (round(corr_mat[i], 3), round(corr_mat_kend[i], 3))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 304,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\t\t(Spearman, Kendal)\n",
      "\n",
      "errors              1.0            1.0\n",
      "av_tok_before_root              -0.103            -0.078\n",
      "CP/T              -0.1            -0.079\n",
      "num_acl              -0.094            -0.075\n",
      "num_adj_noun              -0.084            -0.064\n",
      "num_part_noun              -0.069            -0.057\n",
      "CN/T              -0.064            -0.049\n",
      "min_depth              0.052            0.045\n",
      "T/S              0.049            0.037\n",
      "VP/T              -0.044            -0.033\n",
      "num_sent              -0.043            -0.034\n",
      "C/S              0.043            0.033\n",
      "num_tu              -0.028            -0.021\n",
      "num_noun_inf              -0.025            -0.02\n",
      "max_depth              0.012            0.01\n",
      "av_depth              0.002            0.002\n"
     ]
    }
   ],
   "source": [
    "print(\"\\t\\t(Spearman, Kendal)\\n\")\n",
    "\n",
    "for col, val in sorted(res_dict.items(), key=lambda x: -np.abs(x[1][0])):\n",
    "    print(col,\"            \", val[0], \"          \", val[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 305,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Colname:  Spearman p-value:   Kendall p-value:\n",
      "av_depth        0.9558        0.9489\n",
      "max_depth        0.7841        0.7823\n",
      "min_depth        0.242        0.2404\n",
      "num_acl        0.0356        0.0343\n",
      "num_sent        0.3408        0.34\n",
      "av_tok_before_root        0.0215        0.0209\n",
      "num_tu        0.5299        0.5394\n",
      "num_adj_noun        0.0592        0.0638\n",
      "num_part_noun        0.1244        0.1229\n",
      "num_noun_inf        0.5789        0.5783\n",
      "C/S        0.3312        0.3246\n",
      "VP/T        0.3225        0.3252\n",
      "T/S        0.277        0.2834\n",
      "CP/T        0.0244        0.0208\n",
      "CN/T        0.1541        0.1496\n",
      "errors        0.0        0.0\n"
     ]
    }
   ],
   "source": [
    "from scipy import stats\n",
    "\n",
    "print(\"Colname:  Spearman p-value:   Kendall p-value:\")\n",
    "for col in filtered_data.columns:\n",
    "    val = filtered_data[col].values\n",
    "    print(col, \"      \", round(stats.spearmanr(val, filtered_data[\"errors\"].values)[1], 4), \"      \", \n",
    "    round(stats.kendalltau(val, filtered_data[\"errors\"].values)[1], 4))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}