Compare_models.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "6c4e4a84",
   "metadata": {},
   "source": [
    "# <div style=\"text-align: right;font-family:Times New Roman\">Приложение Б-3</div>\n",
    "# <div style=\"text-align: center;font-family:Times New Roman\">Сравнение моделей</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "524e1fc3",
   "metadata": {},
   "source": [
    "# Model comparison\n",
    "# Сравнение моделей\n",
    "---\n",
    "In this notebook different models - logistic regression, SVM, random forest, gradient boosting, dense neural network and U-Net - are applied to model the vegetation of 1/20 of a territory and their performances are compared\n",
    "\n",
    "В данном блокноте разные модели - логистическая регрессия, метод опорных векторов, случайный лес, градиентный бустинг, полносвязная нейронная сеть и U-Net - применяются для моделирования растительности на 1/20 части территории и сравнивается их качество"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2628d6fe",
   "metadata": {},
   "source": [
    "### Importing dependencies\n",
    "### Импорт необходимых библиотек"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "later-mercury",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\ProgramData\\Anaconda3\\envs\\geo\\lib\\site-packages\\xgboost\\compat.py:36: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.\n",
      "  from pandas import MultiIndex, Int64Index\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import random\n",
    "import re\n",
    "import numpy as np\n",
    "import matplotlib\n",
    "from glob import glob\n",
    "import rasterio as rio\n",
    "import matplotlib.pyplot as plt\n",
    "from matplotlib import patches as mpatches, colors\n",
    "import pandas as pd\n",
    "import pickle\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.svm import LinearSVC\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.metrics import accuracy_score, mean_squared_error, roc_auc_score, f1_score\n",
    "from osgeo import ogr, osr\n",
    "import xgboost as xgb\n",
    "\n",
    "import fiona\n",
    "import rasterio as rio\n",
    "from rasterio.warp import calculate_default_transform, reproject, Resampling\n",
    "import rasterio.merge\n",
    "import rasterio.fill\n",
    "from rasterio.io import MemoryFile\n",
    "from rasterio.enums import Resampling\n",
    "from rasterio.windows import Window\n",
    "from rasterio.transform import Affine\n",
    "import earthpy as et\n",
    "import earthpy.spatial as es\n",
    "import earthpy.plot as ep\n",
    "import earthpy.mask as em\n",
    "\n",
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "from tensorflow.keras import utils\n",
    "from tensorflow.keras import layers\n",
    "from tensorflow.keras import optimizers"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e91618d6",
   "metadata": {},
   "source": [
    "### Data preparation\n",
    "### Подготовка данных"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "planned-integration",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['F:\\\\Work\\\\CorrProizv\\\\final_noadj\\\\B1.tif',\n",
       " 'F:\\\\Work\\\\CorrProizv\\\\final_noadj\\\\B11.tif',\n",
       " 'F:\\\\Work\\\\CorrProizv\\\\final_noadj\\\\B12.tif',\n",
       " 'F:\\\\Work\\\\CorrProizv\\\\final_noadj\\\\B2.tif',\n",
       " 'F:\\\\Work\\\\CorrProizv\\\\final_noadj\\\\B3.tif',\n",
       " 'F:\\\\Work\\\\CorrProizv\\\\final_noadj\\\\B4.tif',\n",
       " 'F:\\\\Work\\\\CorrProizv\\\\final_noadj\\\\B5.tif',\n",
       " 'F:\\\\Work\\\\CorrProizv\\\\final_noadj\\\\B6.tif',\n",
       " 'F:\\\\Work\\\\CorrProizv\\\\final_noadj\\\\B7.tif',\n",
       " 'F:\\\\Work\\\\CorrProizv\\\\final_noadj\\\\B8.tif',\n",
       " 'F:\\\\Work\\\\CorrProizv\\\\final_noadj\\\\B8A.tif',\n",
       " 'F:\\\\Work\\\\CorrProizv\\\\final_noadj\\\\B9.tif']"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#loading x data / загрузка предикторов\n",
    "path = r'F:\\Work\\CorrProizv\\final_noadj\\\\'\n",
    "files = glob(path+'*.tif')\n",
    "files"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "controlled-resident",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "F:\\Work\\CorrProizv\\final_noadj\\B1.tif\n",
      "F:\\Work\\CorrProizv\\final_noadj\\B11.tif\n",
      "F:\\Work\\CorrProizv\\final_noadj\\B12.tif\n",
      "F:\\Work\\CorrProizv\\final_noadj\\B2.tif\n",
      "F:\\Work\\CorrProizv\\final_noadj\\B3.tif\n",
      "F:\\Work\\CorrProizv\\final_noadj\\B4.tif\n",
      "F:\\Work\\CorrProizv\\final_noadj\\B5.tif\n",
      "F:\\Work\\CorrProizv\\final_noadj\\B6.tif\n",
      "F:\\Work\\CorrProizv\\final_noadj\\B7.tif\n",
      "F:\\Work\\CorrProizv\\final_noadj\\B8.tif\n",
      "F:\\Work\\CorrProizv\\final_noadj\\B8A.tif\n",
      "F:\\Work\\CorrProizv\\final_noadj\\B9.tif\n"
     ]
    }
   ],
   "source": [
    "x_train = []\n",
    "for band in files:\n",
    "    with rio.open(band) as bnd:\n",
    "        rstr = bnd.read(1)\n",
    "        x_train.append(rstr[img > 1])\n",
    "    print(band)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "authentic-google",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[array([103, 116, 125, ..., 225, 226, 244], dtype=uint16),\n",
       " array([ 276,  306,  347, ..., 1104, 1099, 1430], dtype=uint16),\n",
       " array([131, 146, 165, ..., 555, 566, 759], dtype=uint16),\n",
       " array([ 73,  85, 102, ..., 246, 252, 312], dtype=uint16),\n",
       " array([181, 177, 199, ..., 359, 419, 447], dtype=uint16),\n",
       " array([ 98, 109, 123, ..., 211, 222, 293], dtype=uint16),\n",
       " array([259, 274, 258, ..., 569, 608, 741], dtype=uint16),\n",
       " array([ 642,  688,  659, ..., 1720, 1647, 1965], dtype=uint16),\n",
       " array([ 697,  755,  766, ..., 2300, 2179, 2565], dtype=uint16),\n",
       " array([ 631,  679,  680, ..., 2040, 1990, 2554], dtype=uint16),\n",
       " array([ 688,  736,  738, ..., 2495, 2378, 2871], dtype=uint16),\n",
       " array([ 540,  594,  619, ..., 2260, 2257, 2773], dtype=uint16)]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "literary-caution",
   "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>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>9</th>\n",
       "      <th>10</th>\n",
       "      <th>11</th>\n",
       "      <th>12</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>103</td>\n",
       "      <td>276</td>\n",
       "      <td>131</td>\n",
       "      <td>73</td>\n",
       "      <td>181</td>\n",
       "      <td>98</td>\n",
       "      <td>259</td>\n",
       "      <td>642</td>\n",
       "      <td>697</td>\n",
       "      <td>631</td>\n",
       "      <td>688</td>\n",
       "      <td>540</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>116</td>\n",
       "      <td>306</td>\n",
       "      <td>146</td>\n",
       "      <td>85</td>\n",
       "      <td>177</td>\n",
       "      <td>109</td>\n",
       "      <td>274</td>\n",
       "      <td>688</td>\n",
       "      <td>755</td>\n",
       "      <td>679</td>\n",
       "      <td>736</td>\n",
       "      <td>594</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>125</td>\n",
       "      <td>347</td>\n",
       "      <td>165</td>\n",
       "      <td>102</td>\n",
       "      <td>199</td>\n",
       "      <td>123</td>\n",
       "      <td>258</td>\n",
       "      <td>659</td>\n",
       "      <td>766</td>\n",
       "      <td>680</td>\n",
       "      <td>738</td>\n",
       "      <td>619</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>135</td>\n",
       "      <td>350</td>\n",
       "      <td>179</td>\n",
       "      <td>122</td>\n",
       "      <td>205</td>\n",
       "      <td>128</td>\n",
       "      <td>248</td>\n",
       "      <td>627</td>\n",
       "      <td>727</td>\n",
       "      <td>685</td>\n",
       "      <td>709</td>\n",
       "      <td>618</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>157</td>\n",
       "      <td>373</td>\n",
       "      <td>184</td>\n",
       "      <td>189</td>\n",
       "      <td>264</td>\n",
       "      <td>143</td>\n",
       "      <td>302</td>\n",
       "      <td>681</td>\n",
       "      <td>708</td>\n",
       "      <td>854</td>\n",
       "      <td>667</td>\n",
       "      <td>665</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>95096420</th>\n",
       "      <td>210</td>\n",
       "      <td>1125</td>\n",
       "      <td>547</td>\n",
       "      <td>239</td>\n",
       "      <td>394</td>\n",
       "      <td>193</td>\n",
       "      <td>641</td>\n",
       "      <td>1762</td>\n",
       "      <td>2278</td>\n",
       "      <td>2224</td>\n",
       "      <td>2593</td>\n",
       "      <td>2409</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>95096421</th>\n",
       "      <td>188</td>\n",
       "      <td>972</td>\n",
       "      <td>451</td>\n",
       "      <td>204</td>\n",
       "      <td>389</td>\n",
       "      <td>188</td>\n",
       "      <td>561</td>\n",
       "      <td>1502</td>\n",
       "      <td>1891</td>\n",
       "      <td>1882</td>\n",
       "      <td>2180</td>\n",
       "      <td>2049</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>95096422</th>\n",
       "      <td>225</td>\n",
       "      <td>1104</td>\n",
       "      <td>555</td>\n",
       "      <td>246</td>\n",
       "      <td>359</td>\n",
       "      <td>211</td>\n",
       "      <td>569</td>\n",
       "      <td>1720</td>\n",
       "      <td>2300</td>\n",
       "      <td>2040</td>\n",
       "      <td>2495</td>\n",
       "      <td>2260</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>95096423</th>\n",
       "      <td>226</td>\n",
       "      <td>1099</td>\n",
       "      <td>566</td>\n",
       "      <td>252</td>\n",
       "      <td>419</td>\n",
       "      <td>222</td>\n",
       "      <td>608</td>\n",
       "      <td>1647</td>\n",
       "      <td>2179</td>\n",
       "      <td>1990</td>\n",
       "      <td>2378</td>\n",
       "      <td>2257</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>95096424</th>\n",
       "      <td>244</td>\n",
       "      <td>1430</td>\n",
       "      <td>759</td>\n",
       "      <td>312</td>\n",
       "      <td>447</td>\n",
       "      <td>293</td>\n",
       "      <td>741</td>\n",
       "      <td>1965</td>\n",
       "      <td>2565</td>\n",
       "      <td>2554</td>\n",
       "      <td>2871</td>\n",
       "      <td>2773</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>95096425 rows × 12 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            1     2    3    4    5    6    7     8     9    10    11    12\n",
       "0         103   276  131   73  181   98  259   642   697   631   688   540\n",
       "1         116   306  146   85  177  109  274   688   755   679   736   594\n",
       "2         125   347  165  102  199  123  258   659   766   680   738   619\n",
       "3         135   350  179  122  205  128  248   627   727   685   709   618\n",
       "4         157   373  184  189  264  143  302   681   708   854   667   665\n",
       "...       ...   ...  ...  ...  ...  ...  ...   ...   ...   ...   ...   ...\n",
       "95096420  210  1125  547  239  394  193  641  1762  2278  2224  2593  2409\n",
       "95096421  188   972  451  204  389  188  561  1502  1891  1882  2180  2049\n",
       "95096422  225  1104  555  246  359  211  569  1720  2300  2040  2495  2260\n",
       "95096423  226  1099  566  252  419  222  608  1647  2179  1990  2378  2257\n",
       "95096424  244  1430  759  312  447  293  741  1965  2565  2554  2871  2773\n",
       "\n",
       "[95096425 rows x 12 columns]"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#converting to pandas dataframe / конвертация в датафрейм pandas\n",
    "x_train = pd.DataFrame(data = {'1': x_train[0], '2': x_train[1], '3': x_train[2], '4': x_train[3], '5': x_train[4], \n",
    "                               '6': x_train[5], '7': x_train[6], '8': x_train[7], '9': x_train[8], '10': x_train[9],\n",
    "                               '11': x_train[10], '12': x_train[11]})\n",
    "x_train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "useful-effectiveness",
   "metadata": {},
   "outputs": [],
   "source": [
    "#loading y data / загрузка обучающих данных\n",
    "y = r'E:\\Work\\CorrProizv\\y_1.tif'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "heavy-movement",
   "metadata": {},
   "outputs": [],
   "source": [
    "with rio.open(y) as bnd:\n",
    "    img = bnd.read(1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "accessory-finland",
   "metadata": {},
   "outputs": [],
   "source": [
    "y_train = img[img > 1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "damaged-success",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([4, 4, 4, ..., 4, 3, 3], dtype=uint16)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "explicit-dialogue",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0           4\n",
       "1           4\n",
       "2           4\n",
       "3           4\n",
       "4           4\n",
       "           ..\n",
       "95096420    3\n",
       "95096421    3\n",
       "95096422    4\n",
       "95096423    3\n",
       "95096424    3\n",
       "Length: 95096425, dtype: uint16"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#converting to pandas series / конвертация в серию pandas\n",
    "y_train = pd.Series(data=y_train)\n",
    "y_train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "large-picnic",
   "metadata": {},
   "outputs": [],
   "source": [
    "img = None\n",
    "rstr = None"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2456c57a",
   "metadata": {},
   "source": [
    "### Logit / Логистическая регрессия"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "25f0e5ad",
   "metadata": {},
   "outputs": [],
   "source": [
    "logit = LogisticRegression()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "ef348613",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\ProgramData\\Anaconda3\\envs\\geo\\lib\\site-packages\\sklearn\\linear_model\\_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "LogisticRegression()"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "logit.fit(x_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "f1bae1f0",
   "metadata": {},
   "outputs": [],
   "source": [
    "x_pred = logit.predict(x_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "85ff812f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.6816483795263597"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy_score(y_train, x_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "0b9bb56a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.4737200793825846"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mean_squared_error(y_train, x_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "23cf179b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.645516229675525"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "roc_auc_score(tf.keras.utils.to_categorical(y_train - 2, num_classes=3), tf.keras.utils.to_categorical(x_pred - 2, num_classes=3))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "12ee62f4",
   "metadata": {},
   "source": [
    "### SVM (Support Vector Machine) / Метод опорных векторов"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "3b0a098a",
   "metadata": {},
   "outputs": [],
   "source": [
    "svm = LinearSVC(max_iter = 10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "93087a72",
   "metadata": {},
   "outputs": [],
   "source": [
    "svm.fit(x_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fe966f2c",
   "metadata": {},
   "outputs": [],
   "source": [
    "x_pred = svm.predict(x_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "28ef8920",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.801137829572957\n"
     ]
    }
   ],
   "source": [
    "accuracy_score(y_train, x_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "f86daa71",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.316284675896952\n"
     ]
    }
   ],
   "source": [
    "mean_squared_error(y_train, x_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "1c716036",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.781659738842042\n"
     ]
    }
   ],
   "source": [
    "roc_auc_score(tf.keras.utils.to_categorical(y_train - 2, num_classes=3), tf.keras.utils.to_categorical(x_pred - 2, num_classes=3))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "dd807cdc",
   "metadata": {},
   "source": [
    "### Random forest / Случайный лес"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0c469ae6",
   "metadata": {},
   "outputs": [],
   "source": [
    "rf = RandomForestClassifier()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "2fae5409",
   "metadata": {},
   "outputs": [],
   "source": [
    "rf.fit(x_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f5711444",
   "metadata": {},
   "outputs": [],
   "source": [
    "x_pred = rf.predict(x_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "65919eee",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.82244932943182\n"
     ]
    }
   ],
   "source": [
    "accuracy_score(y_train, x_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "a5bdfdd6",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.32014956936104\n"
     ]
    }
   ],
   "source": [
    "mean_squared_error(y_train, x_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "588fe4a9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.796430568324985\n"
     ]
    }
   ],
   "source": [
    "roc_auc_score(tf.keras.utils.to_categorical(y_train - 2, num_classes=3), tf.keras.utils.to_categorical(x_pred - 2, num_classes=3))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c7716fd6",
   "metadata": {},
   "source": [
    "### XGboost / Градиентный бустинг"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "ceramic-qualification",
   "metadata": {},
   "outputs": [],
   "source": [
    "xgbc = xgb.XGBClassifier()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "spare-paintball",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\ProgramData\\Anaconda3\\lib\\site-packages\\xgboost\\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[13:49:07] WARNING: D:\\bld\\xgboost-split_1631904903843\\work\\src\\learner.cc:1095: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'multi:softprob' was changed from 'merror' to 'mlogloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "XGBClassifier(base_score=0.5, booster='gbtree', colsample_bylevel=1,\n",
       "              colsample_bynode=1, colsample_bytree=1, gamma=0, gpu_id=-1,\n",
       "              importance_type='gain', interaction_constraints='',\n",
       "              learning_rate=0.300000012, max_delta_step=0, max_depth=6,\n",
       "              min_child_weight=1, missing=nan, monotone_constraints='()',\n",
       "              n_estimators=100, n_jobs=16, num_parallel_tree=1,\n",
       "              objective='multi:softprob', random_state=0, reg_alpha=0,\n",
       "              reg_lambda=1, scale_pos_weight=None, subsample=1,\n",
       "              tree_method='approx', validate_parameters=1, verbosity=None)"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "xgbc.fit(x_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "ruled-mitchell",
   "metadata": {},
   "outputs": [],
   "source": [
    "#saving model / сохранение модели\n",
    "pickle.dump(xgbc, open(\"xgbc.pickle.dat\", \"wb\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "animal-category",
   "metadata": {},
   "outputs": [],
   "source": [
    "xgbc.save_model('xgbc1.json')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "danish-sussex",
   "metadata": {},
   "outputs": [],
   "source": [
    "x_pred = xgbc.predict(x_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "id": "cloudy-number",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8466998988485912"
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy_score(y_train, x_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "nearby-mainstream",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.27254565378138085"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mean_squared_error(y_train, x_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "id": "e6680221",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8466998988485912"
      ]
     },
     "execution_count": 76,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "f1_score(y_train, x_pred, average='micro')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "id": "6e0b94d9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8216102231842349"
      ]
     },
     "execution_count": 68,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "roc_auc_score(keras.utils.to_categorical(y_train - 2, num_classes=3), keras.utils.to_categorical(x_pred - 2, num_classes=3))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9e525791",
   "metadata": {},
   "source": [
    "### Dense neural network (perceptron) / Полносвязная нейронная сеть (перцептрон)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "d1cfbaca",
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow import keras"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "b53857b4",
   "metadata": {},
   "outputs": [],
   "source": [
    "#converting y data to keras categorical format / конвертация обучающих данных в категориальный формат keras\n",
    "y_train = y_train - 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "a5d5b901",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([2, 2, 2, ..., 2, 1, 1], dtype=uint16)"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "b2ab720d",
   "metadata": {},
   "outputs": [],
   "source": [
    "y_train = keras.utils.to_categorical(\n",
    "    y_train, num_classes=3\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "10cb1aed",
   "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>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>9</th>\n",
       "      <th>10</th>\n",
       "      <th>11</th>\n",
       "      <th>12</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>131</td>\n",
       "      <td>1294</td>\n",
       "      <td>610</td>\n",
       "      <td>165</td>\n",
       "      <td>393</td>\n",
       "      <td>211</td>\n",
       "      <td>788</td>\n",
       "      <td>2138</td>\n",
       "      <td>2382</td>\n",
       "      <td>2480</td>\n",
       "      <td>2532</td>\n",
       "      <td>2579</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>121</td>\n",
       "      <td>1385</td>\n",
       "      <td>645</td>\n",
       "      <td>139</td>\n",
       "      <td>459</td>\n",
       "      <td>275</td>\n",
       "      <td>866</td>\n",
       "      <td>2322</td>\n",
       "      <td>2588</td>\n",
       "      <td>2724</td>\n",
       "      <td>2705</td>\n",
       "      <td>2788</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>122</td>\n",
       "      <td>1427</td>\n",
       "      <td>669</td>\n",
       "      <td>152</td>\n",
       "      <td>441</td>\n",
       "      <td>259</td>\n",
       "      <td>865</td>\n",
       "      <td>2356</td>\n",
       "      <td>2609</td>\n",
       "      <td>2784</td>\n",
       "      <td>2881</td>\n",
       "      <td>2891</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>138</td>\n",
       "      <td>1417</td>\n",
       "      <td>668</td>\n",
       "      <td>173</td>\n",
       "      <td>466</td>\n",
       "      <td>266</td>\n",
       "      <td>880</td>\n",
       "      <td>2297</td>\n",
       "      <td>2535</td>\n",
       "      <td>2670</td>\n",
       "      <td>2843</td>\n",
       "      <td>2835</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>146</td>\n",
       "      <td>1435</td>\n",
       "      <td>671</td>\n",
       "      <td>176</td>\n",
       "      <td>450</td>\n",
       "      <td>283</td>\n",
       "      <td>888</td>\n",
       "      <td>2114</td>\n",
       "      <td>2433</td>\n",
       "      <td>2592</td>\n",
       "      <td>2718</td>\n",
       "      <td>2737</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>95094199</th>\n",
       "      <td>210</td>\n",
       "      <td>1125</td>\n",
       "      <td>547</td>\n",
       "      <td>239</td>\n",
       "      <td>394</td>\n",
       "      <td>193</td>\n",
       "      <td>641</td>\n",
       "      <td>1762</td>\n",
       "      <td>2278</td>\n",
       "      <td>2224</td>\n",
       "      <td>2593</td>\n",
       "      <td>2409</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>95094200</th>\n",
       "      <td>188</td>\n",
       "      <td>972</td>\n",
       "      <td>451</td>\n",
       "      <td>204</td>\n",
       "      <td>389</td>\n",
       "      <td>188</td>\n",
       "      <td>561</td>\n",
       "      <td>1502</td>\n",
       "      <td>1891</td>\n",
       "      <td>1882</td>\n",
       "      <td>2180</td>\n",
       "      <td>2049</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>95094201</th>\n",
       "      <td>225</td>\n",
       "      <td>1104</td>\n",
       "      <td>555</td>\n",
       "      <td>246</td>\n",
       "      <td>359</td>\n",
       "      <td>211</td>\n",
       "      <td>569</td>\n",
       "      <td>1720</td>\n",
       "      <td>2300</td>\n",
       "      <td>2040</td>\n",
       "      <td>2495</td>\n",
       "      <td>2260</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>95094202</th>\n",
       "      <td>226</td>\n",
       "      <td>1099</td>\n",
       "      <td>566</td>\n",
       "      <td>252</td>\n",
       "      <td>419</td>\n",
       "      <td>222</td>\n",
       "      <td>608</td>\n",
       "      <td>1647</td>\n",
       "      <td>2179</td>\n",
       "      <td>1990</td>\n",
       "      <td>2378</td>\n",
       "      <td>2257</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>95094203</th>\n",
       "      <td>244</td>\n",
       "      <td>1430</td>\n",
       "      <td>759</td>\n",
       "      <td>312</td>\n",
       "      <td>447</td>\n",
       "      <td>293</td>\n",
       "      <td>741</td>\n",
       "      <td>1965</td>\n",
       "      <td>2565</td>\n",
       "      <td>2554</td>\n",
       "      <td>2871</td>\n",
       "      <td>2773</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>95094204 rows × 12 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            1     2    3    4    5    6    7     8     9    10    11    12\n",
       "0         131  1294  610  165  393  211  788  2138  2382  2480  2532  2579\n",
       "1         121  1385  645  139  459  275  866  2322  2588  2724  2705  2788\n",
       "2         122  1427  669  152  441  259  865  2356  2609  2784  2881  2891\n",
       "3         138  1417  668  173  466  266  880  2297  2535  2670  2843  2835\n",
       "4         146  1435  671  176  450  283  888  2114  2433  2592  2718  2737\n",
       "...       ...   ...  ...  ...  ...  ...  ...   ...   ...   ...   ...   ...\n",
       "95094199  210  1125  547  239  394  193  641  1762  2278  2224  2593  2409\n",
       "95094200  188   972  451  204  389  188  561  1502  1891  1882  2180  2049\n",
       "95094201  225  1104  555  246  359  211  569  1720  2300  2040  2495  2260\n",
       "95094202  226  1099  566  252  419  222  608  1647  2179  1990  2378  2257\n",
       "95094203  244  1430  759  312  447  293  741  1965  2565  2554  2871  2773\n",
       "\n",
       "[95094204 rows x 12 columns]"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "6ff8ddb2",
   "metadata": {},
   "outputs": [],
   "source": [
    "#converting x data to numpy array / конвертация предикторов в массив numpy\n",
    "x_train = x_train.to_numpy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "f07af719",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 131, 1294,  610, ..., 2480, 2532, 2579],\n",
       "       [ 121, 1385,  645, ..., 2724, 2705, 2788],\n",
       "       [ 122, 1427,  669, ..., 2784, 2881, 2891],\n",
       "       ...,\n",
       "       [ 225, 1104,  555, ..., 2040, 2495, 2260],\n",
       "       [ 226, 1099,  566, ..., 1990, 2378, 2257],\n",
       "       [ 244, 1430,  759, ..., 2554, 2871, 2773]], dtype=uint16)"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "102f258b",
   "metadata": {},
   "outputs": [],
   "source": [
    "#random shuffling data and splitting to train and test\n",
    "#случайное перемешивание данных и разбиение на тренировочную и тестовую выборки\n",
    "samples = list(range(len(y_train)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "21f3bd29",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[66921040,\n",
       " 42159898,\n",
       " 88408867,\n",
       " 58351929,\n",
       " 59967678,\n",
       " 19389264,\n",
       " 30268879,\n",
       " 60247734,\n",
       " 63833411,\n",
       " 42109713,\n",
       " 85891440,\n",
       " 13579409,\n",
       " 89382946,\n",
       " 39677134,\n",
       " 1911718,\n",
       " 72143436,\n",
       " 5273515,\n",
       " 32204433,\n",
       " 29138745,\n",
       " 93178616,\n",
       " 46541160,\n",
       " 3331142,\n",
       " 10382305,\n",
       " 21844564,\n",
       " 30761233,\n",
       " 91354915,\n",
       " 7191236,\n",
       " 64773929,\n",
       " 68655205,\n",
       " 26555416,\n",
       " 70442730,\n",
       " 30281104,\n",
       " 86055393,\n",
       " 90599200,\n",
       " 38514638,\n",
       " 31199386,\n",
       " 6090882,\n",
       " 58474481,\n",
       " 18018011,\n",
       " 43571650,\n",
       " 52226241,\n",
       " 21943262,\n",
       " 11981878,\n",
       " 53002301,\n",
       " 8534974,\n",
       " 27380906,\n",
       " 66475979,\n",
       " 49247827,\n",
       " 14698339,\n",
       " 4211532,\n",
       " 79430582,\n",
       " 24506282,\n",
       " 36031035,\n",
       " 69603938,\n",
       " 58604924,\n",
       " 36656070,\n",
       " 55421508,\n",
       " 47892969,\n",
       " 60954128,\n",
       " 94382468,\n",
       " 16866757,\n",
       " 30900210,\n",
       " 42660982,\n",
       " 28787134,\n",
       " 20672543,\n",
       " 62948977,\n",
       " 1991493,\n",
       " 26104087,\n",
       " 7526330,\n",
       " 70326687,\n",
       " 82547273,\n",
       " 18895527,\n",
       " 80915552,\n",
       " 32590386,\n",
       " 21964004,\n",
       " 4408938,\n",
       " 898806,\n",
       " 53374080,\n",
       " 85562627,\n",
       " 42287126,\n",
       " 39529214,\n",
       " 46071027,\n",
       " 87375268,\n",
       " 15300331,\n",
       " 64172053,\n",
       " 34093437,\n",
       " 47248801,\n",
       " 28352474,\n",
       " 24889237,\n",
       " 75196018,\n",
       " 36194259,\n",
       " 75598079,\n",
       " 56257258,\n",
       " 82465549,\n",
       " 36685711,\n",
       " 73656077,\n",
       " 79574842,\n",
       " 63306384,\n",
       " 34667629,\n",
       " 60110663,\n",
       " 79435260,\n",
       " 35525740,\n",
       " 56702418,\n",
       " 22346948,\n",
       " 63709949,\n",
       " 66938758,\n",
       " 13276256,\n",
       " 90203110,\n",
       " 12531229,\n",
       " 14906705,\n",
       " 38747191,\n",
       " 88556274,\n",
       " 90373924,\n",
       " 36139475,\n",
       " 5872917,\n",
       " 37458980,\n",
       " 30227333,\n",
       " 85103346,\n",
       " 13501061,\n",
       " 69019611,\n",
       " 55011541,\n",
       " 39020443,\n",
       " 3918660,\n",
       " 7323450,\n",
       " 26671548,\n",
       " 3239518,\n",
       " 56296506,\n",
       " 18322362,\n",
       " 93412296,\n",
       " 47012939,\n",
       " 12404241,\n",
       " 73622306,\n",
       " 93418816,\n",
       " 56527465,\n",
       " 49238209,\n",
       " 79210903,\n",
       " 78123950,\n",
       " 35665133,\n",
       " 35704897,\n",
       " 61190136,\n",
       " 64304526,\n",
       " 7502571,\n",
       " 43493640,\n",
       " 39144532,\n",
       " 16434946,\n",
       " 47376633,\n",
       " 50196378,\n",
       " 23348167,\n",
       " 53032305,\n",
       " 56523661,\n",
       " 7038335,\n",
       " 22986746,\n",
       " 13400608,\n",
       " 24218293,\n",
       " 14539650,\n",
       " 78206696,\n",
       " 38662317,\n",
       " 65596424,\n",
       " 79244342,\n",
       " 19000030,\n",
       " 31371289,\n",
       " 69282729,\n",
       " 19050413,\n",
       " 47775681,\n",
       " 72121858,\n",
       " 22668677,\n",
       " 80888172,\n",
       " 22601299,\n",
       " 92370532,\n",
       " 70736395,\n",
       " 91992396,\n",
       " 92108863,\n",
       " 44496749,\n",
       " 65113524,\n",
       " 77951444,\n",
       " 31371425,\n",
       " 22580005,\n",
       " 64954165,\n",
       " 10225416,\n",
       " 25988894,\n",
       " 33762116,\n",
       " 86791817,\n",
       " 54961545,\n",
       " 35433935,\n",
       " 23503594,\n",
       " 68255858,\n",
       " 69252733,\n",
       " 37117855,\n",
       " 20793900,\n",
       " 54536939,\n",
       " 86786769,\n",
       " 53217492,\n",
       " 81508390,\n",
       " 80065084,\n",
       " 15758346,\n",
       " 60689337,\n",
       " 18652318,\n",
       " 7057861,\n",
       " 53491233,\n",
       " 21133849,\n",
       " 76242963,\n",
       " 12319896,\n",
       " 49696832,\n",
       " 12376405,\n",
       " 11033357,\n",
       " 68948260,\n",
       " 47110331,\n",
       " 70819836,\n",
       " 71780185,\n",
       " 38644581,\n",
       " 33875522,\n",
       " 48725585,\n",
       " 8976856,\n",
       " 88079434,\n",
       " 33123832,\n",
       " 49132425,\n",
       " 43729800,\n",
       " 10356077,\n",
       " 41758265,\n",
       " 49798829,\n",
       " 30839351,\n",
       " 1449776,\n",
       " 41924478,\n",
       " 84006549,\n",
       " 16962184,\n",
       " 32420165,\n",
       " 30807355,\n",
       " 31632830,\n",
       " 69869990,\n",
       " 53350156,\n",
       " 56094233,\n",
       " 36451524,\n",
       " 22039347,\n",
       " 68162724,\n",
       " 74033822,\n",
       " 14771273,\n",
       " 31972006,\n",
       " 40340343,\n",
       " 12854105,\n",
       " 86316726,\n",
       " 38221997,\n",
       " 35990299,\n",
       " 47285844,\n",
       " 33391577,\n",
       " 18976803,\n",
       " 73851833,\n",
       " 26402150,\n",
       " 38166851,\n",
       " 31179375,\n",
       " 73635405,\n",
       " 72333289,\n",
       " 22512537,\n",
       " 66489099,\n",
       " 33973946,\n",
       " 32316770,\n",
       " 86776409,\n",
       " 18981809,\n",
       " 13717544,\n",
       " 44292534,\n",
       " 58713268,\n",
       " 76980859,\n",
       " 34885874,\n",
       " 39287906,\n",
       " 15463660,\n",
       " 70758702,\n",
       " 94983458,\n",
       " 65522912,\n",
       " 84917441,\n",
       " 36030128,\n",
       " 47841325,\n",
       " 10774402,\n",
       " 4981458,\n",
       " 84080916,\n",
       " 79468359,\n",
       " 60044265,\n",
       " 9428702,\n",
       " 13874066,\n",
       " 21611337,\n",
       " 76740904,\n",
       " 14244458,\n",
       " 91681879,\n",
       " 86167725,\n",
       " 18459268,\n",
       " 31013081,\n",
       " 31398648,\n",
       " 66878212,\n",
       " 80851663,\n",
       " 85082655,\n",
       " 4512206,\n",
       " 87494219,\n",
       " 44147468,\n",
       " 14848832,\n",
       " 47861966,\n",
       " 52195941,\n",
       " 65423138,\n",
       " 51168779,\n",
       " 64806202,\n",
       " 17799887,\n",
       " 68099781,\n",
       " 54791716,\n",
       " 54988744,\n",
       " 53923246,\n",
       " 71612140,\n",
       " 53374324,\n",
       " 80357348,\n",
       " 28414081,\n",
       " 9390251,\n",
       " 50995101,\n",
       " 62776533,\n",
       " 53135916,\n",
       " 23990001,\n",
       " 35693364,\n",
       " 44214757,\n",
       " 1307315,\n",
       " 58215523,\n",
       " 80852122,\n",
       " 27778095,\n",
       " 60548100,\n",
       " 7141992,\n",
       " 5510464,\n",
       " 2252146,\n",
       " 22074516,\n",
       " 50773215,\n",
       " 15099506,\n",
       " 68744686,\n",
       " 13680301,\n",
       " 59440264,\n",
       " 4313040,\n",
       " 23661241,\n",
       " 9048608,\n",
       " 22001547,\n",
       " 71472303,\n",
       " 93364750,\n",
       " 38538943,\n",
       " 87073504,\n",
       " 67968226,\n",
       " 39004873,\n",
       " 50044116,\n",
       " 91038786,\n",
       " 65718932,\n",
       " 38210336,\n",
       " 48973018,\n",
       " 20393863,\n",
       " 91487722,\n",
       " 10139903,\n",
       " 68965279,\n",
       " 82990602,\n",
       " 63428517,\n",
       " 84288494,\n",
       " 22825909,\n",
       " 2385486,\n",
       " 39990744,\n",
       " 59226806,\n",
       " 23863217,\n",
       " 31270375,\n",
       " 708340,\n",
       " 76292493,\n",
       " 8109750,\n",
       " 80744247,\n",
       " 9549747,\n",
       " 7335879,\n",
       " 11284604,\n",
       " 77477274,\n",
       " 44647354,\n",
       " 13485858,\n",
       " 44163969,\n",
       " 49267871,\n",
       " 24538695,\n",
       " 38025890,\n",
       " 6962183,\n",
       " 63465480,\n",
       " 61533459,\n",
       " 13433116,\n",
       " 34456731,\n",
       " 19210261,\n",
       " 17714244,\n",
       " 32217973,\n",
       " 1716118,\n",
       " 52530015,\n",
       " 52808425,\n",
       " 3484875,\n",
       " 40809386,\n",
       " 50637342,\n",
       " 27821005,\n",
       " 48431600,\n",
       " 26411629,\n",
       " 37382999,\n",
       " 15676150,\n",
       " 73541273,\n",
       " 35323003,\n",
       " 46113219,\n",
       " 81632607,\n",
       " 61573395,\n",
       " 77338826,\n",
       " 81206629,\n",
       " 21349431,\n",
       " 24410509,\n",
       " 68479652,\n",
       " 84726735,\n",
       " 61626284,\n",
       " 83160790,\n",
       " 71585565,\n",
       " 80204096,\n",
       " 19768125,\n",
       " 81809789,\n",
       " 17790006,\n",
       " 19732977,\n",
       " 59407303,\n",
       " 44214629,\n",
       " 33007978,\n",
       " 4499973,\n",
       " 6308005,\n",
       " 31332460,\n",
       " 64442731,\n",
       " 64634129,\n",
       " 70684975,\n",
       " 86226783,\n",
       " 50805600,\n",
       " 71905928,\n",
       " 83289472,\n",
       " 89095245,\n",
       " 84435552,\n",
       " 79446431,\n",
       " 17020058,\n",
       " 36207609,\n",
       " 22839311,\n",
       " 34684617,\n",
       " 89526853,\n",
       " 32566122,\n",
       " 57581798,\n",
       " 68797170,\n",
       " 73887016,\n",
       " 3720299,\n",
       " 2182932,\n",
       " 30010260,\n",
       " 35643824,\n",
       " 94244323,\n",
       " 94782189,\n",
       " 24638965,\n",
       " 61897044,\n",
       " 76721511,\n",
       " 42916331,\n",
       " 35215903,\n",
       " 89347136,\n",
       " 55726225,\n",
       " 94167947,\n",
       " 19290330,\n",
       " 4740316,\n",
       " 84159452,\n",
       " 53846114,\n",
       " 19816667,\n",
       " 64110643,\n",
       " 7945507,\n",
       " 13938277,\n",
       " 24268984,\n",
       " 70832975,\n",
       " 84292095,\n",
       " 14326042,\n",
       " 15693785,\n",
       " 20521928,\n",
       " 93081808,\n",
       " 88021734,\n",
       " 20212512,\n",
       " 80549547,\n",
       " 89456474,\n",
       " 70066264,\n",
       " 35348238,\n",
       " 76530575,\n",
       " 80229105,\n",
       " 92644885,\n",
       " 19030745,\n",
       " 51740492,\n",
       " 50858807,\n",
       " 59993429,\n",
       " 39240453,\n",
       " 20914515,\n",
       " 66752885,\n",
       " 57909226,\n",
       " 4609567,\n",
       " 13889968,\n",
       " 29762646,\n",
       " 22859102,\n",
       " 30645826,\n",
       " 81729509,\n",
       " 35972821,\n",
       " 1087480,\n",
       " 68133655,\n",
       " 68177893,\n",
       " 23291859,\n",
       " 10166216,\n",
       " 43299957,\n",
       " 77208826,\n",
       " 23892281,\n",
       " 21254105,\n",
       " 60807337,\n",
       " 53668177,\n",
       " 23238725,\n",
       " 10197589,\n",
       " 82629725,\n",
       " 28845870,\n",
       " 1242202,\n",
       " 48444143,\n",
       " 68895681,\n",
       " 22238878,\n",
       " 51584775,\n",
       " 82251322,\n",
       " 30362091,\n",
       " 82162115,\n",
       " 88599685,\n",
       " 48337260,\n",
       " 53320746,\n",
       " 56041039,\n",
       " 64268301,\n",
       " 3200147,\n",
       " 5849783,\n",
       " 49864035,\n",
       " 29331271,\n",
       " 59975776,\n",
       " 26180997,\n",
       " 21030131,\n",
       " 44341051,\n",
       " 94023079,\n",
       " 11011819,\n",
       " 64027414,\n",
       " 27083503,\n",
       " 21486528,\n",
       " 20691174,\n",
       " 90197383,\n",
       " 70269864,\n",
       " 6855176,\n",
       " 13899267,\n",
       " 40042137,\n",
       " 87937757,\n",
       " 39409418,\n",
       " 69526264,\n",
       " 65034850,\n",
       " 82129343,\n",
       " 12998230,\n",
       " 23749314,\n",
       " 58747562,\n",
       " 27426293,\n",
       " 43869423,\n",
       " 42933245,\n",
       " 54688979,\n",
       " 37763438,\n",
       " 73207556,\n",
       " 89649016,\n",
       " 84582007,\n",
       " 76932410,\n",
       " 85209342,\n",
       " 62829485,\n",
       " 24687420,\n",
       " 49815281,\n",
       " 22110548,\n",
       " 8683081,\n",
       " 25695236,\n",
       " 1838314,\n",
       " 69182967,\n",
       " 46669473,\n",
       " 79611420,\n",
       " 53554134,\n",
       " 25318336,\n",
       " 50666786,\n",
       " 26995416,\n",
       " 35911063,\n",
       " 82666672,\n",
       " 49114661,\n",
       " 25888377,\n",
       " 58870193,\n",
       " 77516082,\n",
       " 89160351,\n",
       " 70577638,\n",
       " 62398269,\n",
       " 35497092,\n",
       " 79408632,\n",
       " 12200824,\n",
       " 20259331,\n",
       " 53438346,\n",
       " 38821403,\n",
       " 56603121,\n",
       " 10706974,\n",
       " 15726209,\n",
       " 87021286,\n",
       " 92163126,\n",
       " 60648090,\n",
       " 59158899,\n",
       " 94472555,\n",
       " 91616759,\n",
       " 12608224,\n",
       " 14353772,\n",
       " 43098255,\n",
       " 35435583,\n",
       " 39593131,\n",
       " 11860339,\n",
       " 57827673,\n",
       " 74035326,\n",
       " 56768891,\n",
       " 7266363,\n",
       " 72545984,\n",
       " 10094684,\n",
       " 18173950,\n",
       " 14588823,\n",
       " 6820430,\n",
       " 29916557,\n",
       " 47723436,\n",
       " 57225051,\n",
       " 35863587,\n",
       " 73320497,\n",
       " 93163535,\n",
       " 48000716,\n",
       " 70300407,\n",
       " 36557234,\n",
       " 87200223,\n",
       " 77575856,\n",
       " 51892206,\n",
       " 22247373,\n",
       " 60949328,\n",
       " 59237905,\n",
       " 55792601,\n",
       " 61612233,\n",
       " 90016750,\n",
       " 89120586,\n",
       " 25284828,\n",
       " 30893831,\n",
       " 71076075,\n",
       " 56563402,\n",
       " 4454154,\n",
       " 27803366,\n",
       " 86803442,\n",
       " 37011220,\n",
       " 36618270,\n",
       " 7423924,\n",
       " 4482601,\n",
       " 93913356,\n",
       " 55547314,\n",
       " 92604769,\n",
       " 49461792,\n",
       " 45160413,\n",
       " 41197372,\n",
       " 82672644,\n",
       " 17559128,\n",
       " 48201066,\n",
       " 24605954,\n",
       " 54425199,\n",
       " 2622834,\n",
       " 43821501,\n",
       " 82694266,\n",
       " 81342002,\n",
       " 87235781,\n",
       " 58115818,\n",
       " 21065769,\n",
       " 93825283,\n",
       " 40904877,\n",
       " 68113376,\n",
       " 43737468,\n",
       " 51867794,\n",
       " 91383539,\n",
       " 3906605,\n",
       " 37072117,\n",
       " 20253627,\n",
       " 65765126,\n",
       " 21724939,\n",
       " 64434168,\n",
       " 26629211,\n",
       " 58101251,\n",
       " 62482771,\n",
       " 84593284,\n",
       " 79086321,\n",
       " 88400861,\n",
       " 10525798,\n",
       " 85167367,\n",
       " 46585457,\n",
       " 1678469,\n",
       " 94697084,\n",
       " 84729050,\n",
       " 11240007,\n",
       " 21919828,\n",
       " 47298661,\n",
       " 48477335,\n",
       " 28305385,\n",
       " 38698057,\n",
       " 57689248,\n",
       " 66868540,\n",
       " 93182421,\n",
       " 31263431,\n",
       " 77643751,\n",
       " 87827115,\n",
       " 40895757,\n",
       " 1943342,\n",
       " 54930279,\n",
       " 61119774,\n",
       " 29166065,\n",
       " 68875691,\n",
       " 16499287,\n",
       " 82902718,\n",
       " 45293196,\n",
       " 59151223,\n",
       " 46688712,\n",
       " 82113205,\n",
       " 91063166,\n",
       " 63534563,\n",
       " 1835673,\n",
       " 25047609,\n",
       " 50628725,\n",
       " 52329354,\n",
       " 52124503,\n",
       " 54898584,\n",
       " 18993763,\n",
       " 43596344,\n",
       " 92755130,\n",
       " 56424645,\n",
       " 60216953,\n",
       " 58234632,\n",
       " 12274080,\n",
       " 81252869,\n",
       " 664810,\n",
       " 63227180,\n",
       " 53872602,\n",
       " 76279399,\n",
       " 22425298,\n",
       " 79067117,\n",
       " 66866851,\n",
       " 24247279,\n",
       " 60898668,\n",
       " 38471455,\n",
       " 43474035,\n",
       " 81033545,\n",
       " 86973706,\n",
       " 24399245,\n",
       " 85119711,\n",
       " 35703975,\n",
       " 12904903,\n",
       " 13818178,\n",
       " 33063072,\n",
       " 65515229,\n",
       " 85308492,\n",
       " 8388055,\n",
       " 15964599,\n",
       " 42222928,\n",
       " 70981283,\n",
       " 59276180,\n",
       " 12649439,\n",
       " 29045291,\n",
       " 3594073,\n",
       " 88172893,\n",
       " 54294550,\n",
       " 28339336,\n",
       " 71723581,\n",
       " 21522874,\n",
       " 6747556,\n",
       " 62914174,\n",
       " 86697757,\n",
       " 38606029,\n",
       " 65036674,\n",
       " 70593586,\n",
       " 5017897,\n",
       " 27264700,\n",
       " 93320924,\n",
       " 81712098,\n",
       " 1500866,\n",
       " 57431121,\n",
       " 45440659,\n",
       " 79978553,\n",
       " 6516251,\n",
       " 24075030,\n",
       " 32586798,\n",
       " 71254191,\n",
       " 81316946,\n",
       " 93212858,\n",
       " 25843179,\n",
       " 24891628,\n",
       " 23737518,\n",
       " 45699608,\n",
       " 69482482,\n",
       " 41932778,\n",
       " 85507365,\n",
       " 34052609,\n",
       " 57737652,\n",
       " 42174971,\n",
       " 15263697,\n",
       " 21477067,\n",
       " 93216248,\n",
       " 3356680,\n",
       " 65809232,\n",
       " 80614341,\n",
       " 60591944,\n",
       " 27629112,\n",
       " 31590455,\n",
       " 83580453,\n",
       " 64535657,\n",
       " 45690106,\n",
       " 65885252,\n",
       " 88060127,\n",
       " 92994879,\n",
       " 47333084,\n",
       " 82077994,\n",
       " 33723394,\n",
       " 69117166,\n",
       " 25419549,\n",
       " 40015152,\n",
       " 50955212,\n",
       " 10716865,\n",
       " 17320887,\n",
       " 83909716,\n",
       " 18307449,\n",
       " 70051196,\n",
       " 75611218,\n",
       " 311842,\n",
       " 79171964,\n",
       " 29429798,\n",
       " 4378825,\n",
       " 73322986,\n",
       " 5483078,\n",
       " 3186162,\n",
       " 54837435,\n",
       " 88207414,\n",
       " 53782241,\n",
       " 21627378,\n",
       " 63084847,\n",
       " 75904646,\n",
       " 61159586,\n",
       " 76013189,\n",
       " 79567770,\n",
       " 29912979,\n",
       " 18588707,\n",
       " 62000827,\n",
       " 52003684,\n",
       " 83875905,\n",
       " 2051965,\n",
       " 4203430,\n",
       " 4810272,\n",
       " 91782051,\n",
       " 22605863,\n",
       " 13905103,\n",
       " 80712476,\n",
       " 59213990,\n",
       " 48231808,\n",
       " 9010420,\n",
       " 60120358,\n",
       " 59180058,\n",
       " 38480926,\n",
       " 3761394,\n",
       " 23488958,\n",
       " 93724086,\n",
       " 33316799,\n",
       " 42868521,\n",
       " 41537957,\n",
       " 26723539,\n",
       " 92928472,\n",
       " 80780519,\n",
       " 47044314,\n",
       " 77462424,\n",
       " 82344188,\n",
       " 69504064,\n",
       " 92754075,\n",
       " 92190495,\n",
       " 51002559,\n",
       " 41424476,\n",
       " 60741517,\n",
       " 38523045,\n",
       " 36919066,\n",
       " 73598559,\n",
       " 81654780,\n",
       " 89539819,\n",
       " 87282054,\n",
       " 49860429,\n",
       " 38481118,\n",
       " 20062405,\n",
       " 5619842,\n",
       " 52270256,\n",
       " 36575481,\n",
       " 5124747,\n",
       " 46715693,\n",
       " 5590210,\n",
       " 52440338,\n",
       " 23605517,\n",
       " 23041013,\n",
       " 63648548,\n",
       " 38924286,\n",
       " 47640999,\n",
       " 9538537,\n",
       " 24083166,\n",
       " 53732792,\n",
       " 69264899,\n",
       " 51737061,\n",
       " 40232746,\n",
       " 73110204,\n",
       " 92130340,\n",
       " 57280559,\n",
       " 42574862,\n",
       " 72090591,\n",
       " 78231632,\n",
       " 34821725,\n",
       " 48326755,\n",
       " 81885558,\n",
       " 93911346,\n",
       " 14635804,\n",
       " 65558193,\n",
       " 32756197,\n",
       " 68256898,\n",
       " 76603441,\n",
       " 577627,\n",
       " 89926782,\n",
       " 27973722,\n",
       " 32315155,\n",
       " 86700153,\n",
       " 14617133,\n",
       " 93548586,\n",
       " 50361161,\n",
       " 87673161,\n",
       " 21258008,\n",
       " 53518818,\n",
       " 64460972,\n",
       " 46212200,\n",
       " 74584261,\n",
       " 60684049,\n",
       " 84603909,\n",
       " 74525588,\n",
       " 24932845,\n",
       " 26885895,\n",
       " 93528578,\n",
       " 86664303,\n",
       " 82893724,\n",
       " 86701788,\n",
       " 19159203,\n",
       " 55196679,\n",
       " 43033364,\n",
       " 61415389,\n",
       " 91893217,\n",
       " 37601782,\n",
       " 50490857,\n",
       " 46149245,\n",
       " 85531171,\n",
       " 83611981,\n",
       " 80633097,\n",
       " 90683899,\n",
       " 87908471,\n",
       " 87813514,\n",
       " 339268,\n",
       " 9245604,\n",
       " 62637472,\n",
       " 84024479,\n",
       " 24266935,\n",
       " 39517813,\n",
       " 59908814,\n",
       " 16483756,\n",
       " 77408171,\n",
       " 73513111,\n",
       " 6210579,\n",
       " 48069622,\n",
       " 90105303,\n",
       " 3717412,\n",
       " 92102667,\n",
       " 25471138,\n",
       " 34596781,\n",
       " 55267139,\n",
       " 52308646,\n",
       " 79744104,\n",
       " 30054638,\n",
       " 33163307,\n",
       " 22623017,\n",
       " 35021066,\n",
       " 88528580,\n",
       " 16925579,\n",
       " 47232085,\n",
       " 24384915,\n",
       " 38208194,\n",
       " 57647716,\n",
       " 73447728,\n",
       " 27647018,\n",
       " 80009402,\n",
       " 67397539,\n",
       " 22894110,\n",
       " 28369324,\n",
       " 89293886,\n",
       " 68252476,\n",
       " 78051510,\n",
       " 20952127,\n",
       " 57026679,\n",
       " 58597740,\n",
       " 5303179,\n",
       " 22726332,\n",
       " 17249184,\n",
       " 20701314,\n",
       " 85087453,\n",
       " 19691481,\n",
       " 34180476,\n",
       " 70186698,\n",
       " 8698397,\n",
       " 88660573,\n",
       " 22771193,\n",
       " 45974130,\n",
       " 79459081,\n",
       " 82229971,\n",
       " 60069507,\n",
       " 27808870,\n",
       " 2640279,\n",
       " 40746914,\n",
       " 62811138,\n",
       " 34567431,\n",
       " ...]"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.random.shuffle(samples)\n",
    "samples"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "d5a64be8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "95094204"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(samples)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "96d2e3e5",
   "metadata": {},
   "outputs": [],
   "source": [
    "train = samples[:80000000]\n",
    "test = samples[80000000:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "75272d18",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 430, 2806, 1586, ..., 3526, 3817, 3842],\n",
       "       [ 251,  245,   96, ...,  405,  260,  607],\n",
       "       [ 312, 1371,  577, ..., 3164, 3293, 3304],\n",
       "       ...,\n",
       "       [ 279, 2435, 1110, ..., 5440, 5698, 5642],\n",
       "       [ 153, 1343,  671, ..., 3752, 3858, 3893],\n",
       "       [ 353,   84,   71, ...,  307,  279,  128]], dtype=uint16)"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_test = x_train[test]\n",
    "y_test = y_train[test]\n",
    "x_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "60d754c4",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "80000000"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_tr = x_train[train]\n",
    "y_tr = y_train[train]\n",
    "len(y_tr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "6a8a2a58",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(80000000, 12)"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_tr.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "98f1049d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "15094204"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(x_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "9c2ca992",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[1., 0., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 1., 0.],\n",
       "       ...,\n",
       "       [1., 0., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [1., 0., 0.]], dtype=float32)"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "d0728f3e",
   "metadata": {},
   "outputs": [],
   "source": [
    "#creating model with 4 dense layers / создание модели с 4 полносвязными слоями\n",
    "model = keras.models.Sequential()\n",
    "model.add(keras.layers.Dense(64, activation='relu', input_shape=(12,)))\n",
    "model.add(keras.layers.Dense(32, activation='relu'))\n",
    "model.add(keras.layers.Dense(16, activation='relu'))\n",
    "model.add(keras.layers.Dense(3, activation='softmax'))\n",
    "model.compile(optimizer='rmsprop', loss='categorical_crossentropy', metrics = ['accuracy', 'mse'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "d5e01565",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense (Dense)                (None, 64)                832       \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 32)                2080      \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 16)                528       \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 3)                 51        \n",
      "=================================================================\n",
      "Total params: 3,491\n",
      "Trainable params: 3,491\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "e873fefc",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5\n",
      "156250/156250 [==============================] - 968s 6ms/step - loss: 0.5092 - accuracy: 0.8223 - mse: 0.0869 - val_loss: 0.4469 - val_accuracy: 0.8385 - val_mse: 0.0796\n",
      "Epoch 2/5\n",
      "156250/156250 [==============================] - 958s 6ms/step - loss: 0.4629 - accuracy: 0.8326 - mse: 0.0822 - val_loss: 0.4554 - val_accuracy: 0.8390 - val_mse: 0.0805\n",
      "Epoch 3/5\n",
      "156250/156250 [==============================] - 964s 6ms/step - loss: 0.4643 - accuracy: 0.8327 - mse: 0.0819 - val_loss: 0.4489 - val_accuracy: 0.8356 - val_mse: 0.0798\n",
      "Epoch 4/5\n",
      "156250/156250 [==============================] - 963s 6ms/step - loss: 0.4540 - accuracy: 0.8350 - mse: 0.0808 - val_loss: 0.4544 - val_accuracy: 0.8378 - val_mse: 0.0801\n",
      "Epoch 5/5\n",
      "156250/156250 [==============================] - 950s 6ms/step - loss: 0.4516 - accuracy: 0.8368 - mse: 0.0803 - val_loss: 0.4517 - val_accuracy: 0.8367 - val_mse: 0.0806\n"
     ]
    }
   ],
   "source": [
    "#training / обучение\n",
    "history = model.fit(x_tr, y_tr, epochs = 5, batch_size = 512, validation_data = (x_test, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "id": "72cc9428",
   "metadata": {},
   "outputs": [],
   "source": [
    "pred_perc = model.predict(x_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "id": "52540beb",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 0, 1, ..., 0, 0, 0], dtype=int64)"
      ]
     },
     "execution_count": 82,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.argmax(pred_perc,axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "id": "df890d79",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8367407781158914"
      ]
     },
     "execution_count": 84,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "f1_score(np.argmax(y_test,axis=1), np.argmax(pred_perc,axis=1), average='micro')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "id": "3190562c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9216162440301868"
      ]
     },
     "execution_count": 69,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "roc_auc_score(y_test, model.predict(x_test))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "667ab92d",
   "metadata": {},
   "source": [
    "### U-NET"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "5929167d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['D:\\\\Work\\\\CorrProizv\\\\1.tif',\n",
       " 'D:\\\\Work\\\\CorrProizv\\\\2.tif',\n",
       " 'D:\\\\Work\\\\CorrProizv\\\\3.tif',\n",
       " 'D:\\\\Work\\\\CorrProizv\\\\4.tif',\n",
       " 'D:\\\\Work\\\\CorrProizv\\\\5.tif',\n",
       " 'D:\\\\Work\\\\CorrProizv\\\\6.tif',\n",
       " 'D:\\\\Work\\\\CorrProizv\\\\7.tif',\n",
       " 'D:\\\\Work\\\\CorrProizv\\\\8.tif',\n",
       " 'D:\\\\Work\\\\CorrProizv\\\\9.tif',\n",
       " 'D:\\\\Work\\\\CorrProizv\\\\10.tif',\n",
       " 'D:\\\\Work\\\\CorrProizv\\\\11.tif',\n",
       " 'D:\\\\Work\\\\CorrProizv\\\\12.tif']"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#importing data / импорт даннных\n",
    "path = r'D:\\Work\\CorrProizv\\\\'\n",
    "files = glob(path+'[0-9].tif') + glob(path+'[0-9][0-9].tif')\n",
    "files"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "e1a89829",
   "metadata": {},
   "outputs": [],
   "source": [
    "y = r'D:\\Work\\CorrProizv\\y.tif'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "b6a264f9",
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train_conv = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "3f5b2d4e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "F:\\Work\\CorrProizv\\final_noadj\\B1.tif\n",
      "F:\\Work\\CorrProizv\\final_noadj\\B11.tif\n",
      "F:\\Work\\CorrProizv\\final_noadj\\B12.tif\n",
      "F:\\Work\\CorrProizv\\final_noadj\\B2.tif\n",
      "F:\\Work\\CorrProizv\\final_noadj\\B3.tif\n",
      "F:\\Work\\CorrProizv\\final_noadj\\B4.tif\n",
      "F:\\Work\\CorrProizv\\final_noadj\\B5.tif\n",
      "F:\\Work\\CorrProizv\\final_noadj\\B6.tif\n",
      "F:\\Work\\CorrProizv\\final_noadj\\B7.tif\n",
      "F:\\Work\\CorrProizv\\final_noadj\\B8.tif\n",
      "F:\\Work\\CorrProizv\\final_noadj\\B8A.tif\n",
      "F:\\Work\\CorrProizv\\final_noadj\\B9.tif\n",
      "F:\\Work\\CorrProizv\\DEM.tif\n",
      "F:\\Work\\CorrProizv\\Landcover.tif\n",
      "F:\\Work\\CorrProizv\\Hansen_mosaic.tif\n"
     ]
    }
   ],
   "source": [
    "for i in range(len(files)):\n",
    "    with rio.open(files[i]) as bnd:\n",
    "        x_train_conv.append(bnd)\n",
    "        print(files[i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "36d7641c",
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train_conv1 = np.stack(x_train_conv, axis = -1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "id": "285e5427",
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train_conv1 = np.transpose(x_train_conv1, axes=[3, 0, 1, 2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "e24404c7",
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train_conv1 = x_train_conv1[0, :, :, :]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "019603fb",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(30613, 38128, 12)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train_conv1.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "b31cc12d",
   "metadata": {},
   "outputs": [],
   "source": [
    "with rio.open(y) as bnd:\n",
    "    y_train_conv = bnd.read().astype('uint16')\n",
    "    final_trans = bnd.transform\n",
    "    #y_train_conv, final_trans = rio.mask.mask(bnd, shape,crop=True, filled=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "3c552e63",
   "metadata": {},
   "outputs": [],
   "source": [
    "#generating tiles / генерация тайлов\n",
    "tiles = []\n",
    "x1 = 0\n",
    "y1 = 0\n",
    "x2 = 128\n",
    "y2 = 128\n",
    "for i in range (0,632,1):\n",
    "    for i in range(0,1089,1):\n",
    "        tiles.append((x1,y1,x2,y2))\n",
    "        y1 +=128\n",
    "        y2 +=128\n",
    "    y1=0\n",
    "    y2=128\n",
    "    x1+=128\n",
    "    x2+=128"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "4972e8be",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(0, 0, 128, 128),\n",
       " (0, 128, 128, 256),\n",
       " (0, 256, 128, 384),\n",
       " (0, 384, 128, 512),\n",
       " (0, 512, 128, 640),\n",
       " (0, 640, 128, 768),\n",
       " (0, 768, 128, 896),\n",
       " (0, 896, 128, 1024),\n",
       " (0, 1024, 128, 1152),\n",
       " (0, 1152, 128, 1280),\n",
       " (0, 1280, 128, 1408),\n",
       " (0, 1408, 128, 1536),\n",
       " (0, 1536, 128, 1664),\n",
       " (0, 1664, 128, 1792),\n",
       " (0, 1792, 128, 1920),\n",
       " (0, 1920, 128, 2048),\n",
       " (0, 2048, 128, 2176),\n",
       " (0, 2176, 128, 2304),\n",
       " (0, 2304, 128, 2432),\n",
       " (0, 2432, 128, 2560),\n",
       " (0, 2560, 128, 2688),\n",
       " (0, 2688, 128, 2816),\n",
       " (0, 2816, 128, 2944),\n",
       " (0, 2944, 128, 3072),\n",
       " (0, 3072, 128, 3200),\n",
       " (0, 3200, 128, 3328),\n",
       " (0, 3328, 128, 3456),\n",
       " (0, 3456, 128, 3584),\n",
       " (0, 3584, 128, 3712),\n",
       " (0, 3712, 128, 3840),\n",
       " (0, 3840, 128, 3968),\n",
       " (0, 3968, 128, 4096),\n",
       " (0, 4096, 128, 4224),\n",
       " (0, 4224, 128, 4352),\n",
       " (0, 4352, 128, 4480),\n",
       " (0, 4480, 128, 4608),\n",
       " (0, 4608, 128, 4736),\n",
       " (0, 4736, 128, 4864),\n",
       " (0, 4864, 128, 4992),\n",
       " (0, 4992, 128, 5120),\n",
       " (0, 5120, 128, 5248),\n",
       " (0, 5248, 128, 5376),\n",
       " (0, 5376, 128, 5504),\n",
       " (0, 5504, 128, 5632),\n",
       " (0, 5632, 128, 5760),\n",
       " (0, 5760, 128, 5888),\n",
       " (0, 5888, 128, 6016),\n",
       " (0, 6016, 128, 6144),\n",
       " (0, 6144, 128, 6272),\n",
       " (0, 6272, 128, 6400),\n",
       " (0, 6400, 128, 6528),\n",
       " (0, 6528, 128, 6656),\n",
       " (0, 6656, 128, 6784),\n",
       " (0, 6784, 128, 6912),\n",
       " (0, 6912, 128, 7040),\n",
       " (0, 7040, 128, 7168),\n",
       " (0, 7168, 128, 7296),\n",
       " (0, 7296, 128, 7424),\n",
       " (0, 7424, 128, 7552),\n",
       " (0, 7552, 128, 7680),\n",
       " (0, 7680, 128, 7808),\n",
       " (0, 7808, 128, 7936),\n",
       " (0, 7936, 128, 8064),\n",
       " (0, 8064, 128, 8192),\n",
       " (0, 8192, 128, 8320),\n",
       " (0, 8320, 128, 8448),\n",
       " (0, 8448, 128, 8576),\n",
       " (0, 8576, 128, 8704),\n",
       " (0, 8704, 128, 8832),\n",
       " (0, 8832, 128, 8960),\n",
       " (0, 8960, 128, 9088),\n",
       " (0, 9088, 128, 9216),\n",
       " (0, 9216, 128, 9344),\n",
       " (0, 9344, 128, 9472),\n",
       " (0, 9472, 128, 9600),\n",
       " (0, 9600, 128, 9728),\n",
       " (0, 9728, 128, 9856),\n",
       " (0, 9856, 128, 9984),\n",
       " (0, 9984, 128, 10112),\n",
       " (0, 10112, 128, 10240),\n",
       " (0, 10240, 128, 10368),\n",
       " (0, 10368, 128, 10496),\n",
       " (0, 10496, 128, 10624),\n",
       " (0, 10624, 128, 10752),\n",
       " (0, 10752, 128, 10880),\n",
       " (0, 10880, 128, 11008),\n",
       " (0, 11008, 128, 11136),\n",
       " (0, 11136, 128, 11264),\n",
       " (0, 11264, 128, 11392),\n",
       " (0, 11392, 128, 11520),\n",
       " (0, 11520, 128, 11648),\n",
       " (0, 11648, 128, 11776),\n",
       " (0, 11776, 128, 11904),\n",
       " (0, 11904, 128, 12032),\n",
       " (0, 12032, 128, 12160),\n",
       " (0, 12160, 128, 12288),\n",
       " (0, 12288, 128, 12416),\n",
       " (0, 12416, 128, 12544),\n",
       " (0, 12544, 128, 12672),\n",
       " (0, 12672, 128, 12800),\n",
       " (0, 12800, 128, 12928),\n",
       " (0, 12928, 128, 13056),\n",
       " (0, 13056, 128, 13184),\n",
       " (0, 13184, 128, 13312),\n",
       " (0, 13312, 128, 13440),\n",
       " (0, 13440, 128, 13568),\n",
       " (0, 13568, 128, 13696),\n",
       " (0, 13696, 128, 13824),\n",
       " (0, 13824, 128, 13952),\n",
       " (0, 13952, 128, 14080),\n",
       " (0, 14080, 128, 14208),\n",
       " (0, 14208, 128, 14336),\n",
       " (0, 14336, 128, 14464),\n",
       " (0, 14464, 128, 14592),\n",
       " (0, 14592, 128, 14720),\n",
       " (0, 14720, 128, 14848),\n",
       " (0, 14848, 128, 14976),\n",
       " (0, 14976, 128, 15104),\n",
       " (0, 15104, 128, 15232),\n",
       " (0, 15232, 128, 15360),\n",
       " (0, 15360, 128, 15488),\n",
       " (0, 15488, 128, 15616),\n",
       " (0, 15616, 128, 15744),\n",
       " (0, 15744, 128, 15872),\n",
       " (0, 15872, 128, 16000),\n",
       " (0, 16000, 128, 16128),\n",
       " (0, 16128, 128, 16256),\n",
       " (0, 16256, 128, 16384),\n",
       " (0, 16384, 128, 16512),\n",
       " (0, 16512, 128, 16640),\n",
       " (0, 16640, 128, 16768),\n",
       " (0, 16768, 128, 16896),\n",
       " (0, 16896, 128, 17024),\n",
       " (0, 17024, 128, 17152),\n",
       " (0, 17152, 128, 17280),\n",
       " (0, 17280, 128, 17408),\n",
       " (0, 17408, 128, 17536),\n",
       " (0, 17536, 128, 17664),\n",
       " (0, 17664, 128, 17792),\n",
       " (0, 17792, 128, 17920),\n",
       " (0, 17920, 128, 18048),\n",
       " (0, 18048, 128, 18176),\n",
       " (0, 18176, 128, 18304),\n",
       " (0, 18304, 128, 18432),\n",
       " (0, 18432, 128, 18560),\n",
       " (0, 18560, 128, 18688),\n",
       " (0, 18688, 128, 18816),\n",
       " (0, 18816, 128, 18944),\n",
       " (0, 18944, 128, 19072),\n",
       " (0, 19072, 128, 19200),\n",
       " (0, 19200, 128, 19328),\n",
       " (0, 19328, 128, 19456),\n",
       " (0, 19456, 128, 19584),\n",
       " (0, 19584, 128, 19712),\n",
       " (0, 19712, 128, 19840),\n",
       " (0, 19840, 128, 19968),\n",
       " (0, 19968, 128, 20096),\n",
       " (0, 20096, 128, 20224),\n",
       " (0, 20224, 128, 20352),\n",
       " (0, 20352, 128, 20480),\n",
       " (0, 20480, 128, 20608),\n",
       " (0, 20608, 128, 20736),\n",
       " (0, 20736, 128, 20864),\n",
       " (0, 20864, 128, 20992),\n",
       " (0, 20992, 128, 21120),\n",
       " (0, 21120, 128, 21248),\n",
       " (0, 21248, 128, 21376),\n",
       " (0, 21376, 128, 21504),\n",
       " (0, 21504, 128, 21632),\n",
       " (0, 21632, 128, 21760),\n",
       " (0, 21760, 128, 21888),\n",
       " (0, 21888, 128, 22016),\n",
       " (0, 22016, 128, 22144),\n",
       " (0, 22144, 128, 22272),\n",
       " (0, 22272, 128, 22400),\n",
       " (0, 22400, 128, 22528),\n",
       " (0, 22528, 128, 22656),\n",
       " (0, 22656, 128, 22784),\n",
       " (0, 22784, 128, 22912),\n",
       " (0, 22912, 128, 23040),\n",
       " (0, 23040, 128, 23168),\n",
       " (0, 23168, 128, 23296),\n",
       " (0, 23296, 128, 23424),\n",
       " (0, 23424, 128, 23552),\n",
       " (0, 23552, 128, 23680),\n",
       " (0, 23680, 128, 23808),\n",
       " (0, 23808, 128, 23936),\n",
       " (0, 23936, 128, 24064),\n",
       " (0, 24064, 128, 24192),\n",
       " (0, 24192, 128, 24320),\n",
       " (0, 24320, 128, 24448),\n",
       " (0, 24448, 128, 24576),\n",
       " (0, 24576, 128, 24704),\n",
       " (0, 24704, 128, 24832),\n",
       " (0, 24832, 128, 24960),\n",
       " (0, 24960, 128, 25088),\n",
       " (0, 25088, 128, 25216),\n",
       " (0, 25216, 128, 25344),\n",
       " (0, 25344, 128, 25472),\n",
       " (0, 25472, 128, 25600),\n",
       " (0, 25600, 128, 25728),\n",
       " (0, 25728, 128, 25856),\n",
       " (0, 25856, 128, 25984),\n",
       " (0, 25984, 128, 26112),\n",
       " (0, 26112, 128, 26240),\n",
       " (0, 26240, 128, 26368),\n",
       " (0, 26368, 128, 26496),\n",
       " (0, 26496, 128, 26624),\n",
       " (0, 26624, 128, 26752),\n",
       " (0, 26752, 128, 26880),\n",
       " (0, 26880, 128, 27008),\n",
       " (0, 27008, 128, 27136),\n",
       " (0, 27136, 128, 27264),\n",
       " (0, 27264, 128, 27392),\n",
       " (0, 27392, 128, 27520),\n",
       " (0, 27520, 128, 27648),\n",
       " (0, 27648, 128, 27776),\n",
       " (0, 27776, 128, 27904),\n",
       " (0, 27904, 128, 28032),\n",
       " (0, 28032, 128, 28160),\n",
       " (0, 28160, 128, 28288),\n",
       " (0, 28288, 128, 28416),\n",
       " (0, 28416, 128, 28544),\n",
       " (0, 28544, 128, 28672),\n",
       " (0, 28672, 128, 28800),\n",
       " (0, 28800, 128, 28928),\n",
       " (0, 28928, 128, 29056),\n",
       " (0, 29056, 128, 29184),\n",
       " (0, 29184, 128, 29312),\n",
       " (0, 29312, 128, 29440),\n",
       " (0, 29440, 128, 29568),\n",
       " (0, 29568, 128, 29696),\n",
       " (0, 29696, 128, 29824),\n",
       " (0, 29824, 128, 29952),\n",
       " (0, 29952, 128, 30080),\n",
       " (0, 30080, 128, 30208),\n",
       " (0, 30208, 128, 30336),\n",
       " (0, 30336, 128, 30464),\n",
       " (0, 30464, 128, 30592),\n",
       " (0, 30592, 128, 30720),\n",
       " (0, 30720, 128, 30848),\n",
       " (0, 30848, 128, 30976),\n",
       " (0, 30976, 128, 31104),\n",
       " (0, 31104, 128, 31232),\n",
       " (0, 31232, 128, 31360),\n",
       " (0, 31360, 128, 31488),\n",
       " (0, 31488, 128, 31616),\n",
       " (0, 31616, 128, 31744),\n",
       " (0, 31744, 128, 31872),\n",
       " (0, 31872, 128, 32000),\n",
       " (0, 32000, 128, 32128),\n",
       " (0, 32128, 128, 32256),\n",
       " (0, 32256, 128, 32384),\n",
       " (0, 32384, 128, 32512),\n",
       " (0, 32512, 128, 32640),\n",
       " (0, 32640, 128, 32768),\n",
       " (0, 32768, 128, 32896),\n",
       " (0, 32896, 128, 33024),\n",
       " (0, 33024, 128, 33152),\n",
       " (0, 33152, 128, 33280),\n",
       " (0, 33280, 128, 33408),\n",
       " (0, 33408, 128, 33536),\n",
       " (0, 33536, 128, 33664),\n",
       " (0, 33664, 128, 33792),\n",
       " (0, 33792, 128, 33920),\n",
       " (0, 33920, 128, 34048),\n",
       " (0, 34048, 128, 34176),\n",
       " (0, 34176, 128, 34304),\n",
       " (0, 34304, 128, 34432),\n",
       " (0, 34432, 128, 34560),\n",
       " (0, 34560, 128, 34688),\n",
       " (0, 34688, 128, 34816),\n",
       " (0, 34816, 128, 34944),\n",
       " (0, 34944, 128, 35072),\n",
       " (0, 35072, 128, 35200),\n",
       " (0, 35200, 128, 35328),\n",
       " (0, 35328, 128, 35456),\n",
       " (0, 35456, 128, 35584),\n",
       " (0, 35584, 128, 35712),\n",
       " (0, 35712, 128, 35840),\n",
       " (0, 35840, 128, 35968),\n",
       " (0, 35968, 128, 36096),\n",
       " (0, 36096, 128, 36224),\n",
       " (0, 36224, 128, 36352),\n",
       " (0, 36352, 128, 36480),\n",
       " (0, 36480, 128, 36608),\n",
       " (0, 36608, 128, 36736),\n",
       " (0, 36736, 128, 36864),\n",
       " (0, 36864, 128, 36992),\n",
       " (0, 36992, 128, 37120),\n",
       " (0, 37120, 128, 37248),\n",
       " (0, 37248, 128, 37376),\n",
       " (0, 37376, 128, 37504),\n",
       " (0, 37504, 128, 37632),\n",
       " (0, 37632, 128, 37760),\n",
       " (0, 37760, 128, 37888),\n",
       " (0, 37888, 128, 38016),\n",
       " (0, 38016, 128, 38144),\n",
       " (0, 38144, 128, 38272),\n",
       " (0, 38272, 128, 38400),\n",
       " (0, 38400, 128, 38528),\n",
       " (0, 38528, 128, 38656),\n",
       " (0, 38656, 128, 38784),\n",
       " (0, 38784, 128, 38912),\n",
       " (0, 38912, 128, 39040),\n",
       " (0, 39040, 128, 39168),\n",
       " (0, 39168, 128, 39296),\n",
       " (0, 39296, 128, 39424),\n",
       " (0, 39424, 128, 39552),\n",
       " (0, 39552, 128, 39680),\n",
       " (0, 39680, 128, 39808),\n",
       " (0, 39808, 128, 39936),\n",
       " (0, 39936, 128, 40064),\n",
       " (0, 40064, 128, 40192),\n",
       " (0, 40192, 128, 40320),\n",
       " (0, 40320, 128, 40448),\n",
       " (0, 40448, 128, 40576),\n",
       " (0, 40576, 128, 40704),\n",
       " (0, 40704, 128, 40832),\n",
       " (0, 40832, 128, 40960),\n",
       " (0, 40960, 128, 41088),\n",
       " (0, 41088, 128, 41216),\n",
       " (0, 41216, 128, 41344),\n",
       " (0, 41344, 128, 41472),\n",
       " (0, 41472, 128, 41600),\n",
       " (0, 41600, 128, 41728),\n",
       " (0, 41728, 128, 41856),\n",
       " (0, 41856, 128, 41984),\n",
       " (0, 41984, 128, 42112),\n",
       " (0, 42112, 128, 42240),\n",
       " (0, 42240, 128, 42368),\n",
       " (0, 42368, 128, 42496),\n",
       " (0, 42496, 128, 42624),\n",
       " (0, 42624, 128, 42752),\n",
       " (0, 42752, 128, 42880),\n",
       " (0, 42880, 128, 43008),\n",
       " (0, 43008, 128, 43136),\n",
       " (0, 43136, 128, 43264),\n",
       " (0, 43264, 128, 43392),\n",
       " (0, 43392, 128, 43520),\n",
       " (0, 43520, 128, 43648),\n",
       " (0, 43648, 128, 43776),\n",
       " (0, 43776, 128, 43904),\n",
       " (0, 43904, 128, 44032),\n",
       " (0, 44032, 128, 44160),\n",
       " (0, 44160, 128, 44288),\n",
       " (0, 44288, 128, 44416),\n",
       " (0, 44416, 128, 44544),\n",
       " (0, 44544, 128, 44672),\n",
       " (0, 44672, 128, 44800),\n",
       " (0, 44800, 128, 44928),\n",
       " (0, 44928, 128, 45056),\n",
       " (0, 45056, 128, 45184),\n",
       " (0, 45184, 128, 45312),\n",
       " (0, 45312, 128, 45440),\n",
       " (0, 45440, 128, 45568),\n",
       " (0, 45568, 128, 45696),\n",
       " (0, 45696, 128, 45824),\n",
       " (0, 45824, 128, 45952),\n",
       " (0, 45952, 128, 46080),\n",
       " (0, 46080, 128, 46208),\n",
       " (0, 46208, 128, 46336),\n",
       " (0, 46336, 128, 46464),\n",
       " (0, 46464, 128, 46592),\n",
       " (0, 46592, 128, 46720),\n",
       " (0, 46720, 128, 46848),\n",
       " (0, 46848, 128, 46976),\n",
       " (0, 46976, 128, 47104),\n",
       " (0, 47104, 128, 47232),\n",
       " (0, 47232, 128, 47360),\n",
       " (0, 47360, 128, 47488),\n",
       " (0, 47488, 128, 47616),\n",
       " (0, 47616, 128, 47744),\n",
       " (0, 47744, 128, 47872),\n",
       " (0, 47872, 128, 48000),\n",
       " (0, 48000, 128, 48128),\n",
       " (0, 48128, 128, 48256),\n",
       " (0, 48256, 128, 48384),\n",
       " (0, 48384, 128, 48512),\n",
       " (0, 48512, 128, 48640),\n",
       " (0, 48640, 128, 48768),\n",
       " (0, 48768, 128, 48896),\n",
       " (0, 48896, 128, 49024),\n",
       " (0, 49024, 128, 49152),\n",
       " (0, 49152, 128, 49280),\n",
       " (0, 49280, 128, 49408),\n",
       " (0, 49408, 128, 49536),\n",
       " (0, 49536, 128, 49664),\n",
       " (0, 49664, 128, 49792),\n",
       " (0, 49792, 128, 49920),\n",
       " (0, 49920, 128, 50048),\n",
       " (0, 50048, 128, 50176),\n",
       " (0, 50176, 128, 50304),\n",
       " (0, 50304, 128, 50432),\n",
       " (0, 50432, 128, 50560),\n",
       " (0, 50560, 128, 50688),\n",
       " (0, 50688, 128, 50816),\n",
       " (0, 50816, 128, 50944),\n",
       " (0, 50944, 128, 51072),\n",
       " (0, 51072, 128, 51200),\n",
       " (0, 51200, 128, 51328),\n",
       " (0, 51328, 128, 51456),\n",
       " (0, 51456, 128, 51584),\n",
       " (0, 51584, 128, 51712),\n",
       " (0, 51712, 128, 51840),\n",
       " (0, 51840, 128, 51968),\n",
       " (0, 51968, 128, 52096),\n",
       " (0, 52096, 128, 52224),\n",
       " (0, 52224, 128, 52352),\n",
       " (0, 52352, 128, 52480),\n",
       " (0, 52480, 128, 52608),\n",
       " (0, 52608, 128, 52736),\n",
       " (0, 52736, 128, 52864),\n",
       " (0, 52864, 128, 52992),\n",
       " (0, 52992, 128, 53120),\n",
       " (0, 53120, 128, 53248),\n",
       " (0, 53248, 128, 53376),\n",
       " (0, 53376, 128, 53504),\n",
       " (0, 53504, 128, 53632),\n",
       " (0, 53632, 128, 53760),\n",
       " (0, 53760, 128, 53888),\n",
       " (0, 53888, 128, 54016),\n",
       " (0, 54016, 128, 54144),\n",
       " (0, 54144, 128, 54272),\n",
       " (0, 54272, 128, 54400),\n",
       " (0, 54400, 128, 54528),\n",
       " (0, 54528, 128, 54656),\n",
       " (0, 54656, 128, 54784),\n",
       " (0, 54784, 128, 54912),\n",
       " (0, 54912, 128, 55040),\n",
       " (0, 55040, 128, 55168),\n",
       " (0, 55168, 128, 55296),\n",
       " (0, 55296, 128, 55424),\n",
       " (0, 55424, 128, 55552),\n",
       " (0, 55552, 128, 55680),\n",
       " (0, 55680, 128, 55808),\n",
       " (0, 55808, 128, 55936),\n",
       " (0, 55936, 128, 56064),\n",
       " (0, 56064, 128, 56192),\n",
       " (0, 56192, 128, 56320),\n",
       " (0, 56320, 128, 56448),\n",
       " (0, 56448, 128, 56576),\n",
       " (0, 56576, 128, 56704),\n",
       " (0, 56704, 128, 56832),\n",
       " (0, 56832, 128, 56960),\n",
       " (0, 56960, 128, 57088),\n",
       " (0, 57088, 128, 57216),\n",
       " (0, 57216, 128, 57344),\n",
       " (0, 57344, 128, 57472),\n",
       " (0, 57472, 128, 57600),\n",
       " (0, 57600, 128, 57728),\n",
       " (0, 57728, 128, 57856),\n",
       " (0, 57856, 128, 57984),\n",
       " (0, 57984, 128, 58112),\n",
       " (0, 58112, 128, 58240),\n",
       " (0, 58240, 128, 58368),\n",
       " (0, 58368, 128, 58496),\n",
       " (0, 58496, 128, 58624),\n",
       " (0, 58624, 128, 58752),\n",
       " (0, 58752, 128, 58880),\n",
       " (0, 58880, 128, 59008),\n",
       " (0, 59008, 128, 59136),\n",
       " (0, 59136, 128, 59264),\n",
       " (0, 59264, 128, 59392),\n",
       " (0, 59392, 128, 59520),\n",
       " (0, 59520, 128, 59648),\n",
       " (0, 59648, 128, 59776),\n",
       " (0, 59776, 128, 59904),\n",
       " (0, 59904, 128, 60032),\n",
       " (0, 60032, 128, 60160),\n",
       " (0, 60160, 128, 60288),\n",
       " (0, 60288, 128, 60416),\n",
       " (0, 60416, 128, 60544),\n",
       " (0, 60544, 128, 60672),\n",
       " (0, 60672, 128, 60800),\n",
       " (0, 60800, 128, 60928),\n",
       " (0, 60928, 128, 61056),\n",
       " (0, 61056, 128, 61184),\n",
       " (0, 61184, 128, 61312),\n",
       " (0, 61312, 128, 61440),\n",
       " (0, 61440, 128, 61568),\n",
       " (0, 61568, 128, 61696),\n",
       " (0, 61696, 128, 61824),\n",
       " (0, 61824, 128, 61952),\n",
       " (0, 61952, 128, 62080),\n",
       " (0, 62080, 128, 62208),\n",
       " (0, 62208, 128, 62336),\n",
       " (0, 62336, 128, 62464),\n",
       " (0, 62464, 128, 62592),\n",
       " (0, 62592, 128, 62720),\n",
       " (0, 62720, 128, 62848),\n",
       " (0, 62848, 128, 62976),\n",
       " (0, 62976, 128, 63104),\n",
       " (0, 63104, 128, 63232),\n",
       " (0, 63232, 128, 63360),\n",
       " (0, 63360, 128, 63488),\n",
       " (0, 63488, 128, 63616),\n",
       " (0, 63616, 128, 63744),\n",
       " (0, 63744, 128, 63872),\n",
       " (0, 63872, 128, 64000),\n",
       " (0, 64000, 128, 64128),\n",
       " (0, 64128, 128, 64256),\n",
       " (0, 64256, 128, 64384),\n",
       " (0, 64384, 128, 64512),\n",
       " (0, 64512, 128, 64640),\n",
       " (0, 64640, 128, 64768),\n",
       " (0, 64768, 128, 64896),\n",
       " (0, 64896, 128, 65024),\n",
       " (0, 65024, 128, 65152),\n",
       " (0, 65152, 128, 65280),\n",
       " (0, 65280, 128, 65408),\n",
       " (0, 65408, 128, 65536),\n",
       " (0, 65536, 128, 65664),\n",
       " (0, 65664, 128, 65792),\n",
       " (0, 65792, 128, 65920),\n",
       " (0, 65920, 128, 66048),\n",
       " (0, 66048, 128, 66176),\n",
       " (0, 66176, 128, 66304),\n",
       " (0, 66304, 128, 66432),\n",
       " (0, 66432, 128, 66560),\n",
       " (0, 66560, 128, 66688),\n",
       " (0, 66688, 128, 66816),\n",
       " (0, 66816, 128, 66944),\n",
       " (0, 66944, 128, 67072),\n",
       " (0, 67072, 128, 67200),\n",
       " (0, 67200, 128, 67328),\n",
       " (0, 67328, 128, 67456),\n",
       " (0, 67456, 128, 67584),\n",
       " (0, 67584, 128, 67712),\n",
       " (0, 67712, 128, 67840),\n",
       " (0, 67840, 128, 67968),\n",
       " (0, 67968, 128, 68096),\n",
       " (0, 68096, 128, 68224),\n",
       " (0, 68224, 128, 68352),\n",
       " (0, 68352, 128, 68480),\n",
       " (0, 68480, 128, 68608),\n",
       " (0, 68608, 128, 68736),\n",
       " (0, 68736, 128, 68864),\n",
       " (0, 68864, 128, 68992),\n",
       " (0, 68992, 128, 69120),\n",
       " (0, 69120, 128, 69248),\n",
       " (0, 69248, 128, 69376),\n",
       " (0, 69376, 128, 69504),\n",
       " (0, 69504, 128, 69632),\n",
       " (0, 69632, 128, 69760),\n",
       " (0, 69760, 128, 69888),\n",
       " (0, 69888, 128, 70016),\n",
       " (0, 70016, 128, 70144),\n",
       " (0, 70144, 128, 70272),\n",
       " (0, 70272, 128, 70400),\n",
       " (0, 70400, 128, 70528),\n",
       " (0, 70528, 128, 70656),\n",
       " (0, 70656, 128, 70784),\n",
       " (0, 70784, 128, 70912),\n",
       " (0, 70912, 128, 71040),\n",
       " (0, 71040, 128, 71168),\n",
       " (0, 71168, 128, 71296),\n",
       " (0, 71296, 128, 71424),\n",
       " (0, 71424, 128, 71552),\n",
       " (0, 71552, 128, 71680),\n",
       " (0, 71680, 128, 71808),\n",
       " (0, 71808, 128, 71936),\n",
       " (0, 71936, 128, 72064),\n",
       " (0, 72064, 128, 72192),\n",
       " (0, 72192, 128, 72320),\n",
       " (0, 72320, 128, 72448),\n",
       " (0, 72448, 128, 72576),\n",
       " (0, 72576, 128, 72704),\n",
       " (0, 72704, 128, 72832),\n",
       " (0, 72832, 128, 72960),\n",
       " (0, 72960, 128, 73088),\n",
       " (0, 73088, 128, 73216),\n",
       " (0, 73216, 128, 73344),\n",
       " (0, 73344, 128, 73472),\n",
       " (0, 73472, 128, 73600),\n",
       " (0, 73600, 128, 73728),\n",
       " (0, 73728, 128, 73856),\n",
       " (0, 73856, 128, 73984),\n",
       " (0, 73984, 128, 74112),\n",
       " (0, 74112, 128, 74240),\n",
       " (0, 74240, 128, 74368),\n",
       " (0, 74368, 128, 74496),\n",
       " (0, 74496, 128, 74624),\n",
       " (0, 74624, 128, 74752),\n",
       " (0, 74752, 128, 74880),\n",
       " (0, 74880, 128, 75008),\n",
       " (0, 75008, 128, 75136),\n",
       " (0, 75136, 128, 75264),\n",
       " (0, 75264, 128, 75392),\n",
       " (0, 75392, 128, 75520),\n",
       " (0, 75520, 128, 75648),\n",
       " (0, 75648, 128, 75776),\n",
       " (0, 75776, 128, 75904),\n",
       " (0, 75904, 128, 76032),\n",
       " (0, 76032, 128, 76160),\n",
       " (0, 76160, 128, 76288),\n",
       " (0, 76288, 128, 76416),\n",
       " (0, 76416, 128, 76544),\n",
       " (0, 76544, 128, 76672),\n",
       " (0, 76672, 128, 76800),\n",
       " (0, 76800, 128, 76928),\n",
       " (0, 76928, 128, 77056),\n",
       " (0, 77056, 128, 77184),\n",
       " (0, 77184, 128, 77312),\n",
       " (0, 77312, 128, 77440),\n",
       " (0, 77440, 128, 77568),\n",
       " (0, 77568, 128, 77696),\n",
       " (0, 77696, 128, 77824),\n",
       " (0, 77824, 128, 77952),\n",
       " (0, 77952, 128, 78080),\n",
       " (0, 78080, 128, 78208),\n",
       " (0, 78208, 128, 78336),\n",
       " (0, 78336, 128, 78464),\n",
       " (0, 78464, 128, 78592),\n",
       " (0, 78592, 128, 78720),\n",
       " (0, 78720, 128, 78848),\n",
       " (0, 78848, 128, 78976),\n",
       " (0, 78976, 128, 79104),\n",
       " (0, 79104, 128, 79232),\n",
       " (0, 79232, 128, 79360),\n",
       " (0, 79360, 128, 79488),\n",
       " (0, 79488, 128, 79616),\n",
       " (0, 79616, 128, 79744),\n",
       " (0, 79744, 128, 79872),\n",
       " (0, 79872, 128, 80000),\n",
       " (0, 80000, 128, 80128),\n",
       " (0, 80128, 128, 80256),\n",
       " (0, 80256, 128, 80384),\n",
       " (0, 80384, 128, 80512),\n",
       " (0, 80512, 128, 80640),\n",
       " (0, 80640, 128, 80768),\n",
       " (0, 80768, 128, 80896),\n",
       " (0, 80896, 128, 81024),\n",
       " (0, 81024, 128, 81152),\n",
       " (0, 81152, 128, 81280),\n",
       " (0, 81280, 128, 81408),\n",
       " (0, 81408, 128, 81536),\n",
       " (0, 81536, 128, 81664),\n",
       " (0, 81664, 128, 81792),\n",
       " (0, 81792, 128, 81920),\n",
       " (0, 81920, 128, 82048),\n",
       " (0, 82048, 128, 82176),\n",
       " (0, 82176, 128, 82304),\n",
       " (0, 82304, 128, 82432),\n",
       " (0, 82432, 128, 82560),\n",
       " (0, 82560, 128, 82688),\n",
       " (0, 82688, 128, 82816),\n",
       " (0, 82816, 128, 82944),\n",
       " (0, 82944, 128, 83072),\n",
       " (0, 83072, 128, 83200),\n",
       " (0, 83200, 128, 83328),\n",
       " (0, 83328, 128, 83456),\n",
       " (0, 83456, 128, 83584),\n",
       " (0, 83584, 128, 83712),\n",
       " (0, 83712, 128, 83840),\n",
       " (0, 83840, 128, 83968),\n",
       " (0, 83968, 128, 84096),\n",
       " (0, 84096, 128, 84224),\n",
       " (0, 84224, 128, 84352),\n",
       " (0, 84352, 128, 84480),\n",
       " (0, 84480, 128, 84608),\n",
       " (0, 84608, 128, 84736),\n",
       " (0, 84736, 128, 84864),\n",
       " (0, 84864, 128, 84992),\n",
       " (0, 84992, 128, 85120),\n",
       " (0, 85120, 128, 85248),\n",
       " (0, 85248, 128, 85376),\n",
       " (0, 85376, 128, 85504),\n",
       " (0, 85504, 128, 85632),\n",
       " (0, 85632, 128, 85760),\n",
       " (0, 85760, 128, 85888),\n",
       " (0, 85888, 128, 86016),\n",
       " (0, 86016, 128, 86144),\n",
       " (0, 86144, 128, 86272),\n",
       " (0, 86272, 128, 86400),\n",
       " (0, 86400, 128, 86528),\n",
       " (0, 86528, 128, 86656),\n",
       " (0, 86656, 128, 86784),\n",
       " (0, 86784, 128, 86912),\n",
       " (0, 86912, 128, 87040),\n",
       " (0, 87040, 128, 87168),\n",
       " (0, 87168, 128, 87296),\n",
       " (0, 87296, 128, 87424),\n",
       " (0, 87424, 128, 87552),\n",
       " (0, 87552, 128, 87680),\n",
       " (0, 87680, 128, 87808),\n",
       " (0, 87808, 128, 87936),\n",
       " (0, 87936, 128, 88064),\n",
       " (0, 88064, 128, 88192),\n",
       " (0, 88192, 128, 88320),\n",
       " (0, 88320, 128, 88448),\n",
       " (0, 88448, 128, 88576),\n",
       " (0, 88576, 128, 88704),\n",
       " (0, 88704, 128, 88832),\n",
       " (0, 88832, 128, 88960),\n",
       " (0, 88960, 128, 89088),\n",
       " (0, 89088, 128, 89216),\n",
       " (0, 89216, 128, 89344),\n",
       " (0, 89344, 128, 89472),\n",
       " (0, 89472, 128, 89600),\n",
       " (0, 89600, 128, 89728),\n",
       " (0, 89728, 128, 89856),\n",
       " (0, 89856, 128, 89984),\n",
       " (0, 89984, 128, 90112),\n",
       " (0, 90112, 128, 90240),\n",
       " (0, 90240, 128, 90368),\n",
       " (0, 90368, 128, 90496),\n",
       " (0, 90496, 128, 90624),\n",
       " (0, 90624, 128, 90752),\n",
       " (0, 90752, 128, 90880),\n",
       " (0, 90880, 128, 91008),\n",
       " (0, 91008, 128, 91136),\n",
       " (0, 91136, 128, 91264),\n",
       " (0, 91264, 128, 91392),\n",
       " (0, 91392, 128, 91520),\n",
       " (0, 91520, 128, 91648),\n",
       " (0, 91648, 128, 91776),\n",
       " (0, 91776, 128, 91904),\n",
       " (0, 91904, 128, 92032),\n",
       " (0, 92032, 128, 92160),\n",
       " (0, 92160, 128, 92288),\n",
       " (0, 92288, 128, 92416),\n",
       " (0, 92416, 128, 92544),\n",
       " (0, 92544, 128, 92672),\n",
       " (0, 92672, 128, 92800),\n",
       " (0, 92800, 128, 92928),\n",
       " (0, 92928, 128, 93056),\n",
       " (0, 93056, 128, 93184),\n",
       " (0, 93184, 128, 93312),\n",
       " (0, 93312, 128, 93440),\n",
       " (0, 93440, 128, 93568),\n",
       " (0, 93568, 128, 93696),\n",
       " (0, 93696, 128, 93824),\n",
       " (0, 93824, 128, 93952),\n",
       " (0, 93952, 128, 94080),\n",
       " (0, 94080, 128, 94208),\n",
       " (0, 94208, 128, 94336),\n",
       " (0, 94336, 128, 94464),\n",
       " (0, 94464, 128, 94592),\n",
       " (0, 94592, 128, 94720),\n",
       " (0, 94720, 128, 94848),\n",
       " (0, 94848, 128, 94976),\n",
       " (0, 94976, 128, 95104),\n",
       " (0, 95104, 128, 95232),\n",
       " (0, 95232, 128, 95360),\n",
       " (0, 95360, 128, 95488),\n",
       " (0, 95488, 128, 95616),\n",
       " (0, 95616, 128, 95744),\n",
       " (0, 95744, 128, 95872),\n",
       " (0, 95872, 128, 96000),\n",
       " (0, 96000, 128, 96128),\n",
       " (0, 96128, 128, 96256),\n",
       " (0, 96256, 128, 96384),\n",
       " (0, 96384, 128, 96512),\n",
       " (0, 96512, 128, 96640),\n",
       " (0, 96640, 128, 96768),\n",
       " (0, 96768, 128, 96896),\n",
       " (0, 96896, 128, 97024),\n",
       " (0, 97024, 128, 97152),\n",
       " (0, 97152, 128, 97280),\n",
       " (0, 97280, 128, 97408),\n",
       " (0, 97408, 128, 97536),\n",
       " (0, 97536, 128, 97664),\n",
       " (0, 97664, 128, 97792),\n",
       " (0, 97792, 128, 97920),\n",
       " (0, 97920, 128, 98048),\n",
       " (0, 98048, 128, 98176),\n",
       " (0, 98176, 128, 98304),\n",
       " (0, 98304, 128, 98432),\n",
       " (0, 98432, 128, 98560),\n",
       " (0, 98560, 128, 98688),\n",
       " (0, 98688, 128, 98816),\n",
       " (0, 98816, 128, 98944),\n",
       " (0, 98944, 128, 99072),\n",
       " (0, 99072, 128, 99200),\n",
       " (0, 99200, 128, 99328),\n",
       " (0, 99328, 128, 99456),\n",
       " (0, 99456, 128, 99584),\n",
       " (0, 99584, 128, 99712),\n",
       " (0, 99712, 128, 99840),\n",
       " (0, 99840, 128, 99968),\n",
       " (0, 99968, 128, 100096),\n",
       " (0, 100096, 128, 100224),\n",
       " (0, 100224, 128, 100352),\n",
       " (0, 100352, 128, 100480),\n",
       " (0, 100480, 128, 100608),\n",
       " (0, 100608, 128, 100736),\n",
       " (0, 100736, 128, 100864),\n",
       " (0, 100864, 128, 100992),\n",
       " (0, 100992, 128, 101120),\n",
       " (0, 101120, 128, 101248),\n",
       " (0, 101248, 128, 101376),\n",
       " (0, 101376, 128, 101504),\n",
       " (0, 101504, 128, 101632),\n",
       " (0, 101632, 128, 101760),\n",
       " (0, 101760, 128, 101888),\n",
       " (0, 101888, 128, 102016),\n",
       " (0, 102016, 128, 102144),\n",
       " (0, 102144, 128, 102272),\n",
       " (0, 102272, 128, 102400),\n",
       " (0, 102400, 128, 102528),\n",
       " (0, 102528, 128, 102656),\n",
       " (0, 102656, 128, 102784),\n",
       " (0, 102784, 128, 102912),\n",
       " (0, 102912, 128, 103040),\n",
       " (0, 103040, 128, 103168),\n",
       " (0, 103168, 128, 103296),\n",
       " (0, 103296, 128, 103424),\n",
       " (0, 103424, 128, 103552),\n",
       " (0, 103552, 128, 103680),\n",
       " (0, 103680, 128, 103808),\n",
       " (0, 103808, 128, 103936),\n",
       " (0, 103936, 128, 104064),\n",
       " (0, 104064, 128, 104192),\n",
       " (0, 104192, 128, 104320),\n",
       " (0, 104320, 128, 104448),\n",
       " (0, 104448, 128, 104576),\n",
       " (0, 104576, 128, 104704),\n",
       " (0, 104704, 128, 104832),\n",
       " (0, 104832, 128, 104960),\n",
       " (0, 104960, 128, 105088),\n",
       " (0, 105088, 128, 105216),\n",
       " (0, 105216, 128, 105344),\n",
       " (0, 105344, 128, 105472),\n",
       " (0, 105472, 128, 105600),\n",
       " (0, 105600, 128, 105728),\n",
       " (0, 105728, 128, 105856),\n",
       " (0, 105856, 128, 105984),\n",
       " (0, 105984, 128, 106112),\n",
       " (0, 106112, 128, 106240),\n",
       " (0, 106240, 128, 106368),\n",
       " (0, 106368, 128, 106496),\n",
       " (0, 106496, 128, 106624),\n",
       " (0, 106624, 128, 106752),\n",
       " (0, 106752, 128, 106880),\n",
       " (0, 106880, 128, 107008),\n",
       " (0, 107008, 128, 107136),\n",
       " (0, 107136, 128, 107264),\n",
       " (0, 107264, 128, 107392),\n",
       " (0, 107392, 128, 107520),\n",
       " (0, 107520, 128, 107648),\n",
       " (0, 107648, 128, 107776),\n",
       " (0, 107776, 128, 107904),\n",
       " (0, 107904, 128, 108032),\n",
       " (0, 108032, 128, 108160),\n",
       " (0, 108160, 128, 108288),\n",
       " (0, 108288, 128, 108416),\n",
       " (0, 108416, 128, 108544),\n",
       " (0, 108544, 128, 108672),\n",
       " (0, 108672, 128, 108800),\n",
       " (0, 108800, 128, 108928),\n",
       " (0, 108928, 128, 109056),\n",
       " (0, 109056, 128, 109184),\n",
       " (0, 109184, 128, 109312),\n",
       " (0, 109312, 128, 109440),\n",
       " (0, 109440, 128, 109568),\n",
       " (0, 109568, 128, 109696),\n",
       " (0, 109696, 128, 109824),\n",
       " (0, 109824, 128, 109952),\n",
       " (0, 109952, 128, 110080),\n",
       " (0, 110080, 128, 110208),\n",
       " (0, 110208, 128, 110336),\n",
       " (0, 110336, 128, 110464),\n",
       " (0, 110464, 128, 110592),\n",
       " (0, 110592, 128, 110720),\n",
       " (0, 110720, 128, 110848),\n",
       " (0, 110848, 128, 110976),\n",
       " (0, 110976, 128, 111104),\n",
       " (0, 111104, 128, 111232),\n",
       " (0, 111232, 128, 111360),\n",
       " (0, 111360, 128, 111488),\n",
       " (0, 111488, 128, 111616),\n",
       " (0, 111616, 128, 111744),\n",
       " (0, 111744, 128, 111872),\n",
       " (0, 111872, 128, 112000),\n",
       " (0, 112000, 128, 112128),\n",
       " (0, 112128, 128, 112256),\n",
       " (0, 112256, 128, 112384),\n",
       " (0, 112384, 128, 112512),\n",
       " (0, 112512, 128, 112640),\n",
       " (0, 112640, 128, 112768),\n",
       " (0, 112768, 128, 112896),\n",
       " (0, 112896, 128, 113024),\n",
       " (0, 113024, 128, 113152),\n",
       " (0, 113152, 128, 113280),\n",
       " (0, 113280, 128, 113408),\n",
       " (0, 113408, 128, 113536),\n",
       " (0, 113536, 128, 113664),\n",
       " (0, 113664, 128, 113792),\n",
       " (0, 113792, 128, 113920),\n",
       " (0, 113920, 128, 114048),\n",
       " (0, 114048, 128, 114176),\n",
       " (0, 114176, 128, 114304),\n",
       " (0, 114304, 128, 114432),\n",
       " (0, 114432, 128, 114560),\n",
       " (0, 114560, 128, 114688),\n",
       " (0, 114688, 128, 114816),\n",
       " (0, 114816, 128, 114944),\n",
       " (0, 114944, 128, 115072),\n",
       " (0, 115072, 128, 115200),\n",
       " (0, 115200, 128, 115328),\n",
       " (0, 115328, 128, 115456),\n",
       " (0, 115456, 128, 115584),\n",
       " (0, 115584, 128, 115712),\n",
       " (0, 115712, 128, 115840),\n",
       " (0, 115840, 128, 115968),\n",
       " (0, 115968, 128, 116096),\n",
       " (0, 116096, 128, 116224),\n",
       " (0, 116224, 128, 116352),\n",
       " (0, 116352, 128, 116480),\n",
       " (0, 116480, 128, 116608),\n",
       " (0, 116608, 128, 116736),\n",
       " (0, 116736, 128, 116864),\n",
       " (0, 116864, 128, 116992),\n",
       " (0, 116992, 128, 117120),\n",
       " (0, 117120, 128, 117248),\n",
       " (0, 117248, 128, 117376),\n",
       " (0, 117376, 128, 117504),\n",
       " (0, 117504, 128, 117632),\n",
       " (0, 117632, 128, 117760),\n",
       " (0, 117760, 128, 117888),\n",
       " (0, 117888, 128, 118016),\n",
       " (0, 118016, 128, 118144),\n",
       " (0, 118144, 128, 118272),\n",
       " (0, 118272, 128, 118400),\n",
       " (0, 118400, 128, 118528),\n",
       " (0, 118528, 128, 118656),\n",
       " (0, 118656, 128, 118784),\n",
       " (0, 118784, 128, 118912),\n",
       " (0, 118912, 128, 119040),\n",
       " (0, 119040, 128, 119168),\n",
       " (0, 119168, 128, 119296),\n",
       " (0, 119296, 128, 119424),\n",
       " (0, 119424, 128, 119552),\n",
       " (0, 119552, 128, 119680),\n",
       " (0, 119680, 128, 119808),\n",
       " (0, 119808, 128, 119936),\n",
       " (0, 119936, 128, 120064),\n",
       " (0, 120064, 128, 120192),\n",
       " (0, 120192, 128, 120320),\n",
       " (0, 120320, 128, 120448),\n",
       " (0, 120448, 128, 120576),\n",
       " (0, 120576, 128, 120704),\n",
       " (0, 120704, 128, 120832),\n",
       " (0, 120832, 128, 120960),\n",
       " (0, 120960, 128, 121088),\n",
       " (0, 121088, 128, 121216),\n",
       " (0, 121216, 128, 121344),\n",
       " (0, 121344, 128, 121472),\n",
       " (0, 121472, 128, 121600),\n",
       " (0, 121600, 128, 121728),\n",
       " (0, 121728, 128, 121856),\n",
       " (0, 121856, 128, 121984),\n",
       " (0, 121984, 128, 122112),\n",
       " (0, 122112, 128, 122240),\n",
       " (0, 122240, 128, 122368),\n",
       " (0, 122368, 128, 122496),\n",
       " (0, 122496, 128, 122624),\n",
       " (0, 122624, 128, 122752),\n",
       " (0, 122752, 128, 122880),\n",
       " (0, 122880, 128, 123008),\n",
       " (0, 123008, 128, 123136),\n",
       " (0, 123136, 128, 123264),\n",
       " (0, 123264, 128, 123392),\n",
       " (0, 123392, 128, 123520),\n",
       " (0, 123520, 128, 123648),\n",
       " (0, 123648, 128, 123776),\n",
       " (0, 123776, 128, 123904),\n",
       " (0, 123904, 128, 124032),\n",
       " (0, 124032, 128, 124160),\n",
       " (0, 124160, 128, 124288),\n",
       " (0, 124288, 128, 124416),\n",
       " (0, 124416, 128, 124544),\n",
       " (0, 124544, 128, 124672),\n",
       " (0, 124672, 128, 124800),\n",
       " (0, 124800, 128, 124928),\n",
       " (0, 124928, 128, 125056),\n",
       " (0, 125056, 128, 125184),\n",
       " (0, 125184, 128, 125312),\n",
       " (0, 125312, 128, 125440),\n",
       " (0, 125440, 128, 125568),\n",
       " (0, 125568, 128, 125696),\n",
       " (0, 125696, 128, 125824),\n",
       " (0, 125824, 128, 125952),\n",
       " (0, 125952, 128, 126080),\n",
       " (0, 126080, 128, 126208),\n",
       " (0, 126208, 128, 126336),\n",
       " (0, 126336, 128, 126464),\n",
       " (0, 126464, 128, 126592),\n",
       " (0, 126592, 128, 126720),\n",
       " (0, 126720, 128, 126848),\n",
       " (0, 126848, 128, 126976),\n",
       " (0, 126976, 128, 127104),\n",
       " (0, 127104, 128, 127232),\n",
       " (0, 127232, 128, 127360),\n",
       " (0, 127360, 128, 127488),\n",
       " (0, 127488, 128, 127616),\n",
       " (0, 127616, 128, 127744),\n",
       " (0, 127744, 128, 127872),\n",
       " (0, 127872, 128, 128000),\n",
       " ...]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tiles"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "41bd3e16",
   "metadata": {},
   "outputs": [],
   "source": [
    "x_tiles = []\n",
    "for tile in tiles:\n",
    "    tilearr = x_train_conv1[tile[0]:tile[2], tile[1]:tile[3],:]\n",
    "    if tilearr.max() > 0:\n",
    "        x_tiles.append(tilearr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "5187a0e3",
   "metadata": {},
   "outputs": [],
   "source": [
    "x_tiles1 = np.stack(x_tiles, axis = 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "aa3e4859",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(6369, 128, 128, 12)"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_tiles1.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "36975e8c",
   "metadata": {},
   "outputs": [],
   "source": [
    "y_tiles = []\n",
    "for tile in tiles:\n",
    "    tilearr = y_train_conv[tile[0]:tile[2], tile[1]:tile[3],:]\n",
    "    if tilearr.max() > 0:\n",
    "        y_tiles.append(tilearr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "ff69f290",
   "metadata": {},
   "outputs": [],
   "source": [
    "y_tiles1 = np.stack(y_tiles, axis = 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "c7437ce2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(6369, 128, 128, 1)"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_tiles1.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "a6d18472",
   "metadata": {},
   "outputs": [],
   "source": [
    "#splitting to train and test data / разделение на тренировочный и тестовый набор данных\n",
    "x_tiles_train = x_tiles1[[samples[:5000]], :, :, :]\n",
    "x_tiles_train = np.squeeze(x_tiles_train, axis=0)\n",
    "y_tiles_train = y_tiles1[[samples[:5000]], :, :, :]\n",
    "y_tiles_train = np.squeeze(y_tiles_train, axis=0)\n",
    "y_tiles_train = keras.utils.to_categorical(y_tiles_train, num_classes=5)\n",
    "x_tiles_test = x_tiles1[[samples[5000:]], :, :, :]\n",
    "x_tiles_test = np.squeeze(x_tiles_test, axis=0)\n",
    "y_tiles_test = y_tiles1[[samples[5000:]], :, :, :]\n",
    "y_tiles_test = np.squeeze(y_tiles_test, axis=0)\n",
    "y_tiles_test = keras.utils.to_categorical(y_tiles_test, num_classes=5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "id": "d20024f7",
   "metadata": {},
   "outputs": [],
   "source": [
    "#creating a model / создание модели\n",
    "from keras import layers, models\n",
    "inputs = layers.Input(shape = (128, 128, 12))\n",
    "\n",
    "conv1 = layers.Conv2D(64, (3, 3), activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(inputs)\n",
    "conv1 = layers.Conv2D(64, (3, 3), activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(conv1)\n",
    "pool1 = layers.MaxPooling2D(pool_size=(2, 2))(conv1)\n",
    "conv2 = layers.Conv2D(128, (3, 3), activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(pool1)\n",
    "conv2 = layers.Conv2D(128, (3, 3), activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(conv2)\n",
    "pool2 = layers.MaxPooling2D(pool_size=(2, 2))(conv2)\n",
    "conv3 = layers.Conv2D(256, (3, 3), activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(pool2)\n",
    "conv3 = layers.Conv2D(256, (3, 3), activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(conv3)\n",
    "pool3 = layers.MaxPooling2D(pool_size=(2, 2))(conv3)\n",
    "conv4 = layers.Conv2D(512, (3, 3), activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(pool3)\n",
    "conv4 = layers.Conv2D(512, (3, 3), activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(conv4)\n",
    "drop4 = layers.Dropout(0.5)(conv4)\n",
    "pool4 = layers.MaxPooling2D(pool_size=(2, 2))(drop4)\n",
    "\n",
    "conv5 = layers.Conv2D(1024, (3, 3), activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(pool4)\n",
    "conv5 = layers.Conv2D(1024, (3, 3), activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(conv5)\n",
    "drop5 = layers.Dropout(0.5)(conv5)\n",
    "\n",
    "up6 = layers.Conv2D(512, (2, 2), activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(layers.UpSampling2D(size = (2,2))(drop5))\n",
    "merge6 = layers.concatenate([drop4,up6], axis = 3)\n",
    "conv6 = layers.Conv2D(512, (3, 3), activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(merge6)\n",
    "conv6 = layers.Conv2D(512, (3, 3), activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(conv6)\n",
    "\n",
    "up7 = layers.Conv2D(256, (2, 2), activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(layers.UpSampling2D(size = (2,2))(conv6))\n",
    "merge7 = layers.concatenate([conv3,up7], axis = 3)\n",
    "conv7 = layers.Conv2D(256, (3, 3), activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(merge7)\n",
    "conv7 = layers.Conv2D(256, (3, 3), activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(conv7)\n",
    "\n",
    "up8 = layers.Conv2D(128, (2, 2), activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(layers.UpSampling2D(size = (2,2))(conv7))\n",
    "merge8 = layers.concatenate([conv2,up8], axis = 3)\n",
    "conv8 = layers.Conv2D(128, (3, 3), activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(merge8)\n",
    "conv8 = layers.Conv2D(128, (3, 3), activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(conv8)\n",
    "\n",
    "up9 = layers.Conv2D(64, (2, 2), activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(layers.UpSampling2D(size = (2,2))(conv8))\n",
    "merge9 = layers.concatenate([conv1,up9], axis = 3)\n",
    "conv9 = layers.Conv2D(64, (3, 3), activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(merge9)\n",
    "conv9 = layers.Conv2D(64, (3, 3), activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(conv9)\n",
    "#conv9 = layers.Conv2D(2, (3, 3), activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(conv9)\n",
    "#conv10 = layers.Conv2D(1, 1, activation = 'softmax', padding = 'same')(conv9)\n",
    "conv10 = layers.Conv2D(5, (1, 1), activation = 'softmax', padding = 'same')(conv9)\n",
    "\n",
    "model = models.Model(inputs = inputs, outputs = conv10)\n",
    "\n",
    "model.compile(optimizer = keras.optimizers.Adam(learning_rate = 0.001), loss = 'categorical_crossentropy', metrics = ['accuracy', 'mse', 'AUC'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "id": "924bb975",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/25\n",
      "625/625 [==============================] - 163s 261ms/step - loss: 2.9677 - accuracy: 0.6944 - mse: 0.1086 - auc: 0.8693 - val_loss: 1.8307 - val_accuracy: 0.6558 - val_mse: 0.1144 - val_auc: 0.8765\n",
      "Epoch 2/25\n",
      "625/625 [==============================] - 163s 261ms/step - loss: 1.0361 - accuracy: 0.7525 - mse: 0.0780 - auc: 0.9307 - val_loss: 0.6951 - val_accuracy: 0.8000 - val_mse: 0.0602 - val_auc: 0.9563\n",
      "Epoch 3/25\n",
      "625/625 [==============================] - 163s 260ms/step - loss: 0.6310 - accuracy: 0.8023 - mse: 0.0588 - auc: 0.9588 - val_loss: 0.5631 - val_accuracy: 0.8319 - val_mse: 0.0513 - val_auc: 0.9681\n",
      "Epoch 4/25\n",
      "625/625 [==============================] - 163s 261ms/step - loss: 0.5227 - accuracy: 0.8238 - mse: 0.0515 - auc: 0.9675 - val_loss: 0.5182 - val_accuracy: 0.8150 - val_mse: 0.0533 - val_auc: 0.9660\n",
      "Epoch 5/25\n",
      "625/625 [==============================] - 163s 261ms/step - loss: 0.5007 - accuracy: 0.8236 - mse: 0.0509 - auc: 0.9689 - val_loss: 0.4561 - val_accuracy: 0.8405 - val_mse: 0.0461 - val_auc: 0.9747\n",
      "Epoch 6/25\n",
      "625/625 [==============================] - 163s 261ms/step - loss: 0.4640 - accuracy: 0.8336 - mse: 0.0478 - auc: 0.9724 - val_loss: 0.4012 - val_accuracy: 0.8509 - val_mse: 0.0425 - val_auc: 0.9789\n",
      "Epoch 7/25\n",
      "625/625 [==============================] - 163s 261ms/step - loss: 0.4348 - accuracy: 0.8419 - mse: 0.0453 - auc: 0.9753 - val_loss: 0.4447 - val_accuracy: 0.8358 - val_mse: 0.0469 - val_auc: 0.9750\n",
      "Epoch 8/25\n",
      "625/625 [==============================] - 163s 261ms/step - loss: 0.4351 - accuracy: 0.8427 - mse: 0.0454 - auc: 0.9752 - val_loss: 0.4158 - val_accuracy: 0.8461 - val_mse: 0.0442 - val_auc: 0.9773\n",
      "Epoch 9/25\n",
      "625/625 [==============================] - 163s 261ms/step - loss: 0.4288 - accuracy: 0.8450 - mse: 0.0450 - auc: 0.9758 - val_loss: 0.6703 - val_accuracy: 0.7757 - val_mse: 0.0653 - val_auc: 0.9546\n",
      "Epoch 10/25\n",
      "625/625 [==============================] - 163s 261ms/step - loss: 0.4472 - accuracy: 0.8378 - mse: 0.0469 - auc: 0.9738 - val_loss: 0.4187 - val_accuracy: 0.8404 - val_mse: 0.0439 - val_auc: 0.9775\n",
      "Epoch 11/25\n",
      "625/625 [==============================] - 163s 260ms/step - loss: 0.4208 - accuracy: 0.8454 - mse: 0.0441 - auc: 0.9767 - val_loss: 0.5145 - val_accuracy: 0.7878 - val_mse: 0.0571 - val_auc: 0.9672\n",
      "Epoch 12/25\n",
      "625/625 [==============================] - 163s 260ms/step - loss: 0.4351 - accuracy: 0.8406 - mse: 0.0456 - auc: 0.9752 - val_loss: 0.4154 - val_accuracy: 0.8423 - val_mse: 0.0448 - val_auc: 0.9773\n",
      "Epoch 13/25\n",
      "625/625 [==============================] - 162s 260ms/step - loss: 0.4957 - accuracy: 0.8281 - mse: 0.0501 - auc: 0.9697 - val_loss: 0.4778 - val_accuracy: 0.8276 - val_mse: 0.0487 - val_auc: 0.9731\n",
      "Epoch 14/25\n",
      "625/625 [==============================] - 163s 261ms/step - loss: 0.4590 - accuracy: 0.8358 - mse: 0.0475 - auc: 0.9730 - val_loss: 0.4787 - val_accuracy: 0.8221 - val_mse: 0.0512 - val_auc: 0.9719\n",
      "Epoch 15/25\n",
      "625/625 [==============================] - 163s 260ms/step - loss: 0.4536 - accuracy: 0.8356 - mse: 0.0473 - auc: 0.9733 - val_loss: 0.4701 - val_accuracy: 0.8278 - val_mse: 0.0511 - val_auc: 0.9713\n",
      "Epoch 16/25\n",
      "625/625 [==============================] - 163s 260ms/step - loss: 0.4322 - accuracy: 0.8415 - mse: 0.0451 - auc: 0.9757 - val_loss: 0.4510 - val_accuracy: 0.8410 - val_mse: 0.0450 - val_auc: 0.9744\n",
      "Epoch 17/25\n",
      "625/625 [==============================] - 162s 259ms/step - loss: 4.3028 - accuracy: 0.6383 - mse: 0.0932 - auc: 0.9124 - val_loss: 0.8445 - val_accuracy: 0.6106 - val_mse: 0.1009 - val_auc: 0.8905\n",
      "Epoch 18/25\n",
      "625/625 [==============================] - 162s 259ms/step - loss: 0.8456 - accuracy: 0.6149 - mse: 0.1010 - auc: 0.8906 - val_loss: 0.8411 - val_accuracy: 0.6110 - val_mse: 0.1008 - val_auc: 0.8906\n",
      "Epoch 19/25\n",
      "625/625 [==============================] - 162s 259ms/step - loss: 0.8431 - accuracy: 0.6153 - mse: 0.1009 - auc: 0.8898 - val_loss: 0.8407 - val_accuracy: 0.6112 - val_mse: 0.1009 - val_auc: 0.8908\n",
      "Epoch 20/25\n",
      "625/625 [==============================] - 162s 259ms/step - loss: 0.8409 - accuracy: 0.6155 - mse: 0.1008 - auc: 0.8902 - val_loss: 0.8374 - val_accuracy: 0.6116 - val_mse: 0.1007 - val_auc: 0.8912\n",
      "Epoch 21/25\n",
      "625/625 [==============================] - 162s 259ms/step - loss: 0.8384 - accuracy: 0.6159 - mse: 0.1007 - auc: 0.8905 - val_loss: 0.8358 - val_accuracy: 0.6119 - val_mse: 0.1005 - val_auc: 0.8915\n",
      "Epoch 22/25\n",
      "625/625 [==============================] - 162s 259ms/step - loss: 0.8416 - accuracy: 0.6155 - mse: 0.1009 - auc: 0.8904 - val_loss: 0.8413 - val_accuracy: 0.6108 - val_mse: 0.1009 - val_auc: 0.8904\n",
      "Epoch 23/25\n",
      "625/625 [==============================] - 161s 258ms/step - loss: 0.8410 - accuracy: 0.6153 - mse: 0.1009 - auc: 0.8903 - val_loss: 0.8384 - val_accuracy: 0.6114 - val_mse: 0.1009 - val_auc: 0.8910\n",
      "Epoch 24/25\n",
      "625/625 [==============================] - 162s 259ms/step - loss: 0.8388 - accuracy: 0.6158 - mse: 0.1008 - auc: 0.8906 - val_loss: 0.8448 - val_accuracy: 0.6103 - val_mse: 0.1012 - val_auc: 0.8900\n",
      "Epoch 25/25\n",
      "625/625 [==============================] - 162s 259ms/step - loss: 0.8390 - accuracy: 0.6156 - mse: 0.1008 - auc: 0.8897 - val_loss: 0.8337 - val_accuracy: 0.6120 - val_mse: 0.1006 - val_auc: 0.8916\n"
     ]
    }
   ],
   "source": [
    "#training a model / обучение модели\n",
    "history = model.fit(x_tiles_train, y_tiles_train, batch_size = 8, epochs = 25, validation_data = (x_tiles_test, y_tiles_test))"
   ]
  }
 ],
 "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": "3.10.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}