├── .gitignore
└── Logistic regression.ipynb


/.gitignore:
--------------------------------------------------------------------------------
1 | .ipynb_checkpoints
2 | 


--------------------------------------------------------------------------------
/Logistic regression.ipynb:
--------------------------------------------------------------------------------
   1 | {
   2 |  "cells": [
   3 |   {
   4 |    "cell_type": "markdown",
   5 |    "metadata": {},
   6 |    "source": [
   7 |     "# Логистическая регрессия"
   8 |    ]
   9 |   },
  10 |   {
  11 |    "cell_type": "code",
  12 |    "execution_count": 1,
  13 |    "metadata": {},
  14 |    "outputs": [],
  15 |    "source": [
  16 |     "# Логистическая регрессия - метод, который позволяет исследовать взаимосвязи между\n",
  17 |     "# номинативной зависимой переменной (ЗП) и различными предикторами.\n",
  18 |     "\n",
  19 |     "# Уравнение логистической регрессии:"
  20 |    ]
  21 |   },
  22 |   {
  23 |    "cell_type": "markdown",
  24 |    "metadata": {},
  25 |    "source": [
  26 |     "\\begin{equation}\n",
  27 |     "log(odds) = B_0 + B_1 x_1 + B_2 x_2 + ... + B_k x_k\n",
  28 |     "\\end{equation}"
  29 |    ]
  30 |   },
  31 |   {
  32 |    "cell_type": "code",
  33 |    "execution_count": 2,
  34 |    "metadata": {},
  35 |    "outputs": [],
  36 |    "source": [
  37 |     "# log - натуральный логарифм.\n",
  38 |     "# odds - шансы положительного исхода (далее просто \"шансы\")."
  39 |    ]
  40 |   },
  41 |   {
  42 |    "cell_type": "markdown",
  43 |    "metadata": {},
  44 |    "source": [
  45 |     "\\begin{equation}\n",
  46 |     "p = \\frac{e^{log(odds)}}{1 + e^{log(odds)}} = \\frac{exp(log(odds))}{1 + exp(log(odds))}\n",
  47 |     "\\end{equation}"
  48 |    ]
  49 |   },
  50 |   {
  51 |    "cell_type": "code",
  52 |    "execution_count": 3,
  53 |    "metadata": {},
  54 |    "outputs": [],
  55 |    "source": [
  56 |     "# p - вероятность положительного исхода (далее просто \"вероятность\")."
  57 |    ]
  58 |   },
  59 |   {
  60 |    "cell_type": "markdown",
  61 |    "metadata": {},
  62 |    "source": [
  63 |     "## Шансы и вероятность"
  64 |    ]
  65 |   },
  66 |   {
  67 |    "cell_type": "code",
  68 |    "execution_count": 4,
  69 |    "metadata": {},
  70 |    "outputs": [],
  71 |    "source": [
  72 |     "import pandas as pd\n",
  73 |     "import numpy as np\n",
  74 |     "import matplotlib.pyplot as plt\n",
  75 |     "import seaborn as sns\n",
  76 |     "import statsmodels.api as sm\n",
  77 |     "import statsmodels.formula.api as smf"
  78 |    ]
  79 |   },
  80 |   {
  81 |    "cell_type": "code",
  82 |    "execution_count": 5,
  83 |    "metadata": {},
  84 |    "outputs": [
  85 |     {
  86 |      "data": {
  87 |       "text/html": [
  88 |        "<div>\n",
  89 |        "<style scoped>\n",
  90 |        "    .dataframe tbody tr th:only-of-type {\n",
  91 |        "        vertical-align: middle;\n",
  92 |        "    }\n",
  93 |        "\n",
  94 |        "    .dataframe tbody tr th {\n",
  95 |        "        vertical-align: top;\n",
  96 |        "    }\n",
  97 |        "\n",
  98 |        "    .dataframe thead th {\n",
  99 |        "        text-align: right;\n",
 100 |        "    }\n",
 101 |        "</style>\n",
 102 |        "<table border=\"1\" class=\"dataframe\">\n",
 103 |        "  <thead>\n",
 104 |        "    <tr style=\"text-align: right;\">\n",
 105 |        "      <th></th>\n",
 106 |        "      <th>coin_state</th>\n",
 107 |        "    </tr>\n",
 108 |        "  </thead>\n",
 109 |        "  <tbody>\n",
 110 |        "    <tr>\n",
 111 |        "      <td>0</td>\n",
 112 |        "      <td>0</td>\n",
 113 |        "    </tr>\n",
 114 |        "    <tr>\n",
 115 |        "      <td>1</td>\n",
 116 |        "      <td>0</td>\n",
 117 |        "    </tr>\n",
 118 |        "    <tr>\n",
 119 |        "      <td>2</td>\n",
 120 |        "      <td>1</td>\n",
 121 |        "    </tr>\n",
 122 |        "    <tr>\n",
 123 |        "      <td>3</td>\n",
 124 |        "      <td>0</td>\n",
 125 |        "    </tr>\n",
 126 |        "    <tr>\n",
 127 |        "      <td>4</td>\n",
 128 |        "      <td>0</td>\n",
 129 |        "    </tr>\n",
 130 |        "    <tr>\n",
 131 |        "      <td>5</td>\n",
 132 |        "      <td>0</td>\n",
 133 |        "    </tr>\n",
 134 |        "    <tr>\n",
 135 |        "      <td>6</td>\n",
 136 |        "      <td>1</td>\n",
 137 |        "    </tr>\n",
 138 |        "    <tr>\n",
 139 |        "      <td>7</td>\n",
 140 |        "      <td>0</td>\n",
 141 |        "    </tr>\n",
 142 |        "    <tr>\n",
 143 |        "      <td>8</td>\n",
 144 |        "      <td>0</td>\n",
 145 |        "    </tr>\n",
 146 |        "    <tr>\n",
 147 |        "      <td>9</td>\n",
 148 |        "      <td>0</td>\n",
 149 |        "    </tr>\n",
 150 |        "  </tbody>\n",
 151 |        "</table>\n",
 152 |        "</div>"
 153 |       ],
 154 |       "text/plain": [
 155 |        "   coin_state\n",
 156 |        "0           0\n",
 157 |        "1           0\n",
 158 |        "2           1\n",
 159 |        "3           0\n",
 160 |        "4           0\n",
 161 |        "5           0\n",
 162 |        "6           1\n",
 163 |        "7           0\n",
 164 |        "8           0\n",
 165 |        "9           0"
 166 |       ]
 167 |      },
 168 |      "execution_count": 5,
 169 |      "metadata": {},
 170 |      "output_type": "execute_result"
 171 |     }
 172 |    ],
 173 |    "source": [
 174 |     "# Разберем на примере подбрасывания монетки. Пусть у нас было 10 бросков и мы получили следующие результаты:\n",
 175 |     "# 1 - орел (в примере мы хотим, чтобы выпадали орлы, это будет положительный исход)\n",
 176 |     "# 0 - решка (отрицательный исход)\n",
 177 |     "\n",
 178 |     "df = pd.DataFrame({'coin_state': [0, 0, 1, 0, 0, 0, 1, 0, 0, 0]})\n",
 179 |     "df"
 180 |    ]
 181 |   },
 182 |   {
 183 |    "cell_type": "code",
 184 |    "execution_count": 6,
 185 |    "metadata": {},
 186 |    "outputs": [
 187 |     {
 188 |      "data": {
 189 |       "text/plain": [
 190 |        "0    8\n",
 191 |        "1    2\n",
 192 |        "Name: coin_state, dtype: int64"
 193 |       ]
 194 |      },
 195 |      "execution_count": 6,
 196 |      "metadata": {},
 197 |      "output_type": "execute_result"
 198 |     }
 199 |    ],
 200 |    "source": [
 201 |     "# 2 орла, 8 решек\n",
 202 |     "df.coin_state.value_counts()"
 203 |    ]
 204 |   },
 205 |   {
 206 |    "cell_type": "code",
 207 |    "execution_count": 7,
 208 |    "metadata": {},
 209 |    "outputs": [],
 210 |    "source": [
 211 |     "# Рассчитаем вероятность:"
 212 |    ]
 213 |   },
 214 |   {
 215 |    "cell_type": "markdown",
 216 |    "metadata": {},
 217 |    "source": [
 218 |     "\\begin{equation}\n",
 219 |     "p = \\frac{pos}{pos + neg}\n",
 220 |     "\\end{equation}  "
 221 |    ]
 222 |   },
 223 |   {
 224 |    "cell_type": "code",
 225 |    "execution_count": 8,
 226 |    "metadata": {},
 227 |    "outputs": [
 228 |     {
 229 |      "data": {
 230 |       "text/plain": [
 231 |        "0.2"
 232 |       ]
 233 |      },
 234 |      "execution_count": 8,
 235 |      "metadata": {},
 236 |      "output_type": "execute_result"
 237 |     }
 238 |    ],
 239 |    "source": [
 240 |     "pos = df.coin_state.value_counts()[1]   # количество положительных исходов\n",
 241 |     "neg = df.coin_state.value_counts()[0]   # количество отрицательных исходов\n",
 242 |     "\n",
 243 |     "p = pos / (pos + neg)\n",
 244 |     "p"
 245 |    ]
 246 |   },
 247 |   {
 248 |    "cell_type": "code",
 249 |    "execution_count": 9,
 250 |    "metadata": {},
 251 |    "outputs": [],
 252 |    "source": [
 253 |     "# Рассчитаем шансы:"
 254 |    ]
 255 |   },
 256 |   {
 257 |    "cell_type": "markdown",
 258 |    "metadata": {},
 259 |    "source": [
 260 |     "\\begin{equation}\n",
 261 |     "odds = \\frac{pos}{neg}\n",
 262 |     "\\end{equation}"
 263 |    ]
 264 |   },
 265 |   {
 266 |    "cell_type": "code",
 267 |    "execution_count": 10,
 268 |    "metadata": {},
 269 |    "outputs": [
 270 |     {
 271 |      "data": {
 272 |       "text/plain": [
 273 |        "0.25"
 274 |       ]
 275 |      },
 276 |      "execution_count": 10,
 277 |      "metadata": {},
 278 |      "output_type": "execute_result"
 279 |     }
 280 |    ],
 281 |    "source": [
 282 |     "odds = pos / neg\n",
 283 |     "odds"
 284 |    ]
 285 |   },
 286 |   {
 287 |    "cell_type": "code",
 288 |    "execution_count": 11,
 289 |    "metadata": {},
 290 |    "outputs": [],
 291 |    "source": [
 292 |     "# Шансы можно посчитать, зная вероятность:"
 293 |    ]
 294 |   },
 295 |   {
 296 |    "cell_type": "markdown",
 297 |    "metadata": {},
 298 |    "source": [
 299 |     "\\begin{equation}\n",
 300 |     "odds = \\frac{p}{1 - p}\n",
 301 |     "\\end{equation}"
 302 |    ]
 303 |   },
 304 |   {
 305 |    "cell_type": "code",
 306 |    "execution_count": 12,
 307 |    "metadata": {},
 308 |    "outputs": [
 309 |     {
 310 |      "data": {
 311 |       "text/plain": [
 312 |        "0.25"
 313 |       ]
 314 |      },
 315 |      "execution_count": 12,
 316 |      "metadata": {},
 317 |      "output_type": "execute_result"
 318 |     }
 319 |    ],
 320 |    "source": [
 321 |     "p / (1 - p)"
 322 |    ]
 323 |   },
 324 |   {
 325 |    "cell_type": "code",
 326 |    "execution_count": 13,
 327 |    "metadata": {},
 328 |    "outputs": [],
 329 |    "source": [
 330 |     "# Вероятность можно посчитать, зная шансы:"
 331 |    ]
 332 |   },
 333 |   {
 334 |    "cell_type": "markdown",
 335 |    "metadata": {},
 336 |    "source": [
 337 |     "\\begin{equation}\n",
 338 |     "p = \\frac{odds}{1 + odds}\n",
 339 |     "\\end{equation}"
 340 |    ]
 341 |   },
 342 |   {
 343 |    "cell_type": "code",
 344 |    "execution_count": 14,
 345 |    "metadata": {},
 346 |    "outputs": [
 347 |     {
 348 |      "data": {
 349 |       "text/plain": [
 350 |        "0.2"
 351 |       ]
 352 |      },
 353 |      "execution_count": 14,
 354 |      "metadata": {},
 355 |      "output_type": "execute_result"
 356 |     }
 357 |    ],
 358 |    "source": [
 359 |     "odds / (1 + odds)"
 360 |    ]
 361 |   },
 362 |   {
 363 |    "cell_type": "markdown",
 364 |    "metadata": {},
 365 |    "source": [
 366 |     "## При чем тут натуральный логарифм?"
 367 |    ]
 368 |   },
 369 |   {
 370 |    "cell_type": "code",
 371 |    "execution_count": 15,
 372 |    "metadata": {},
 373 |    "outputs": [],
 374 |    "source": [
 375 |     "# Проследим за различным соотношением положительных и отрицательных исходов\n",
 376 |     "# и рассчитанными на их основе вероятностью, шансами и натуральным логарифмом шансов (далее просто \"логарифм шансов\").\n",
 377 |     "\n",
 378 |     "# Вероятность     - число от нуля до единицы\n",
 379 |     "# Шансы           - число от нуля до плюс бесконечности\n",
 380 |     "# Логарифм шансов - число от минус бесконечности до плюс бесконечности"
 381 |    ]
 382 |   },
 383 |   {
 384 |    "cell_type": "code",
 385 |    "execution_count": 16,
 386 |    "metadata": {},
 387 |    "outputs": [
 388 |     {
 389 |      "data": {
 390 |       "text/html": [
 391 |        "<div>\n",
 392 |        "<style scoped>\n",
 393 |        "    .dataframe tbody tr th:only-of-type {\n",
 394 |        "        vertical-align: middle;\n",
 395 |        "    }\n",
 396 |        "\n",
 397 |        "    .dataframe tbody tr th {\n",
 398 |        "        vertical-align: top;\n",
 399 |        "    }\n",
 400 |        "\n",
 401 |        "    .dataframe thead th {\n",
 402 |        "        text-align: right;\n",
 403 |        "    }\n",
 404 |        "</style>\n",
 405 |        "<table border=\"1\" class=\"dataframe\">\n",
 406 |        "  <thead>\n",
 407 |        "    <tr style=\"text-align: right;\">\n",
 408 |        "      <th></th>\n",
 409 |        "      <th>pos</th>\n",
 410 |        "      <th>neg</th>\n",
 411 |        "      <th>p</th>\n",
 412 |        "      <th>odds</th>\n",
 413 |        "      <th>log(odds)</th>\n",
 414 |        "    </tr>\n",
 415 |        "  </thead>\n",
 416 |        "  <tbody>\n",
 417 |        "    <tr>\n",
 418 |        "      <td>0</td>\n",
 419 |        "      <td>0</td>\n",
 420 |        "      <td>10</td>\n",
 421 |        "      <td>0.0</td>\n",
 422 |        "      <td>0.00</td>\n",
 423 |        "      <td>NaN</td>\n",
 424 |        "    </tr>\n",
 425 |        "    <tr>\n",
 426 |        "      <td>1</td>\n",
 427 |        "      <td>1</td>\n",
 428 |        "      <td>0</td>\n",
 429 |        "      <td>0.1</td>\n",
 430 |        "      <td>0.11</td>\n",
 431 |        "      <td>-2.20</td>\n",
 432 |        "    </tr>\n",
 433 |        "    <tr>\n",
 434 |        "      <td>2</td>\n",
 435 |        "      <td>2</td>\n",
 436 |        "      <td>8</td>\n",
 437 |        "      <td>0.2</td>\n",
 438 |        "      <td>0.25</td>\n",
 439 |        "      <td>-1.39</td>\n",
 440 |        "    </tr>\n",
 441 |        "    <tr>\n",
 442 |        "      <td>3</td>\n",
 443 |        "      <td>3</td>\n",
 444 |        "      <td>7</td>\n",
 445 |        "      <td>0.3</td>\n",
 446 |        "      <td>0.43</td>\n",
 447 |        "      <td>-0.85</td>\n",
 448 |        "    </tr>\n",
 449 |        "    <tr>\n",
 450 |        "      <td>4</td>\n",
 451 |        "      <td>4</td>\n",
 452 |        "      <td>6</td>\n",
 453 |        "      <td>0.4</td>\n",
 454 |        "      <td>0.66</td>\n",
 455 |        "      <td>-0.41</td>\n",
 456 |        "    </tr>\n",
 457 |        "    <tr>\n",
 458 |        "      <td>5</td>\n",
 459 |        "      <td>5</td>\n",
 460 |        "      <td>5</td>\n",
 461 |        "      <td>0.5</td>\n",
 462 |        "      <td>1.00</td>\n",
 463 |        "      <td>0.00</td>\n",
 464 |        "    </tr>\n",
 465 |        "    <tr>\n",
 466 |        "      <td>6</td>\n",
 467 |        "      <td>6</td>\n",
 468 |        "      <td>4</td>\n",
 469 |        "      <td>0.6</td>\n",
 470 |        "      <td>1.50</td>\n",
 471 |        "      <td>0.41</td>\n",
 472 |        "    </tr>\n",
 473 |        "    <tr>\n",
 474 |        "      <td>7</td>\n",
 475 |        "      <td>7</td>\n",
 476 |        "      <td>3</td>\n",
 477 |        "      <td>0.7</td>\n",
 478 |        "      <td>2.33</td>\n",
 479 |        "      <td>0.85</td>\n",
 480 |        "    </tr>\n",
 481 |        "    <tr>\n",
 482 |        "      <td>8</td>\n",
 483 |        "      <td>8</td>\n",
 484 |        "      <td>2</td>\n",
 485 |        "      <td>0.8</td>\n",
 486 |        "      <td>4.00</td>\n",
 487 |        "      <td>1.39</td>\n",
 488 |        "    </tr>\n",
 489 |        "    <tr>\n",
 490 |        "      <td>9</td>\n",
 491 |        "      <td>9</td>\n",
 492 |        "      <td>1</td>\n",
 493 |        "      <td>0.9</td>\n",
 494 |        "      <td>9.00</td>\n",
 495 |        "      <td>2.20</td>\n",
 496 |        "    </tr>\n",
 497 |        "    <tr>\n",
 498 |        "      <td>10</td>\n",
 499 |        "      <td>10</td>\n",
 500 |        "      <td>0</td>\n",
 501 |        "      <td>1.0</td>\n",
 502 |        "      <td>NaN</td>\n",
 503 |        "      <td>NaN</td>\n",
 504 |        "    </tr>\n",
 505 |        "  </tbody>\n",
 506 |        "</table>\n",
 507 |        "</div>"
 508 |       ],
 509 |       "text/plain": [
 510 |        "    pos  neg    p  odds  log(odds)\n",
 511 |        "0     0   10  0.0  0.00        NaN\n",
 512 |        "1     1    0  0.1  0.11      -2.20\n",
 513 |        "2     2    8  0.2  0.25      -1.39\n",
 514 |        "3     3    7  0.3  0.43      -0.85\n",
 515 |        "4     4    6  0.4  0.66      -0.41\n",
 516 |        "5     5    5  0.5  1.00       0.00\n",
 517 |        "6     6    4  0.6  1.50       0.41\n",
 518 |        "7     7    3  0.7  2.33       0.85\n",
 519 |        "8     8    2  0.8  4.00       1.39\n",
 520 |        "9     9    1  0.9  9.00       2.20\n",
 521 |        "10   10    0  1.0   NaN        NaN"
 522 |       ]
 523 |      },
 524 |      "execution_count": 16,
 525 |      "metadata": {},
 526 |      "output_type": "execute_result"
 527 |     }
 528 |    ],
 529 |    "source": [
 530 |     "pd.DataFrame({\n",
 531 |     "    'pos': [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10],\n",
 532 |     "    'neg': [10, 0, 8, 7, 6, 5, 4, 3, 2, 1, 0],\n",
 533 |     "    'p': [0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1],\n",
 534 |     "    'odds': [0, 0.11, 0.25, 0.43, 0.66, 1, 1.5, 2.33, 4, 9, np.nan],\n",
 535 |     "    'log(odds)': [np.nan, -2.2, -1.39, -0.85, -0.41, 0, 0.41, 0.85, 1.39, 2.2, np.nan],\n",
 536 |     "})"
 537 |    ]
 538 |   },
 539 |   {
 540 |    "cell_type": "code",
 541 |    "execution_count": 17,
 542 |    "metadata": {},
 543 |    "outputs": [
 544 |     {
 545 |      "data": {
 546 |       "image/png": "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\n",
 547 |       "text/plain": [
 548 |        "<Figure size 432x288 with 1 Axes>"
 549 |       ]
 550 |      },
 551 |      "metadata": {
 552 |       "needs_background": "light"
 553 |      },
 554 |      "output_type": "display_data"
 555 |     }
 556 |    ],
 557 |    "source": [
 558 |     "# Вспомним уравнение логистической регрессии - в правой части число от минус бесконечности до плюс бесконечности,\n",
 559 |     "# в левой - логарифм шансов, зная который, мы можем рассчитать вероятность.\n",
 560 |     "\n",
 561 |     "# Визуализируем зависимость вероятности от логарифма шансов:\n",
 562 |     "\n",
 563 |     "log_odds = np.arange(-6, 6, 0.1)\n",
 564 |     "p = np.exp(log_odds) / (1 + np.exp(log_odds))\n",
 565 |     "\n",
 566 |     "plt.plot(log_odds, p)\n",
 567 |     "plt.grid()\n",
 568 |     "plt.xlim(-6, 6)\n",
 569 |     "plt.ylim(-0.1, 1.1)\n",
 570 |     "plt.title('Зависимость вероятности от логарифма шансов', fontsize=12)\n",
 571 |     "plt.xlabel('log(odds)', fontsize=12)\n",
 572 |     "plt.ylabel('p', fontsize=12)\n",
 573 |     "\n",
 574 |     "plt.show()\n",
 575 |     "plt.close()"
 576 |    ]
 577 |   },
 578 |   {
 579 |    "cell_type": "markdown",
 580 |    "metadata": {},
 581 |    "source": [
 582 |     "## Модель без предикторов"
 583 |    ]
 584 |   },
 585 |   {
 586 |    "cell_type": "code",
 587 |    "execution_count": 18,
 588 |    "metadata": {},
 589 |    "outputs": [
 590 |     {
 591 |      "data": {
 592 |       "text/html": [
 593 |        "<div>\n",
 594 |        "<style scoped>\n",
 595 |        "    .dataframe tbody tr th:only-of-type {\n",
 596 |        "        vertical-align: middle;\n",
 597 |        "    }\n",
 598 |        "\n",
 599 |        "    .dataframe tbody tr th {\n",
 600 |        "        vertical-align: top;\n",
 601 |        "    }\n",
 602 |        "\n",
 603 |        "    .dataframe thead th {\n",
 604 |        "        text-align: right;\n",
 605 |        "    }\n",
 606 |        "</style>\n",
 607 |        "<table border=\"1\" class=\"dataframe\">\n",
 608 |        "  <thead>\n",
 609 |        "    <tr style=\"text-align: right;\">\n",
 610 |        "      <th></th>\n",
 611 |        "      <th>PassengerId</th>\n",
 612 |        "      <th>Survived</th>\n",
 613 |        "      <th>Pclass</th>\n",
 614 |        "      <th>Name</th>\n",
 615 |        "      <th>Sex</th>\n",
 616 |        "      <th>Age</th>\n",
 617 |        "      <th>SibSp</th>\n",
 618 |        "      <th>Parch</th>\n",
 619 |        "      <th>Ticket</th>\n",
 620 |        "      <th>Fare</th>\n",
 621 |        "      <th>Cabin</th>\n",
 622 |        "      <th>Embarked</th>\n",
 623 |        "    </tr>\n",
 624 |        "  </thead>\n",
 625 |        "  <tbody>\n",
 626 |        "    <tr>\n",
 627 |        "      <td>0</td>\n",
 628 |        "      <td>1</td>\n",
 629 |        "      <td>0</td>\n",
 630 |        "      <td>3</td>\n",
 631 |        "      <td>Braund, Mr. Owen Harris</td>\n",
 632 |        "      <td>male</td>\n",
 633 |        "      <td>22.0</td>\n",
 634 |        "      <td>1</td>\n",
 635 |        "      <td>0</td>\n",
 636 |        "      <td>A/5 21171</td>\n",
 637 |        "      <td>7.2500</td>\n",
 638 |        "      <td>NaN</td>\n",
 639 |        "      <td>S</td>\n",
 640 |        "    </tr>\n",
 641 |        "    <tr>\n",
 642 |        "      <td>1</td>\n",
 643 |        "      <td>2</td>\n",
 644 |        "      <td>1</td>\n",
 645 |        "      <td>1</td>\n",
 646 |        "      <td>Cumings, Mrs. John Bradley (Florence Briggs Th...</td>\n",
 647 |        "      <td>female</td>\n",
 648 |        "      <td>38.0</td>\n",
 649 |        "      <td>1</td>\n",
 650 |        "      <td>0</td>\n",
 651 |        "      <td>PC 17599</td>\n",
 652 |        "      <td>71.2833</td>\n",
 653 |        "      <td>C85</td>\n",
 654 |        "      <td>C</td>\n",
 655 |        "    </tr>\n",
 656 |        "    <tr>\n",
 657 |        "      <td>2</td>\n",
 658 |        "      <td>3</td>\n",
 659 |        "      <td>1</td>\n",
 660 |        "      <td>3</td>\n",
 661 |        "      <td>Heikkinen, Miss. Laina</td>\n",
 662 |        "      <td>female</td>\n",
 663 |        "      <td>26.0</td>\n",
 664 |        "      <td>0</td>\n",
 665 |        "      <td>0</td>\n",
 666 |        "      <td>STON/O2. 3101282</td>\n",
 667 |        "      <td>7.9250</td>\n",
 668 |        "      <td>NaN</td>\n",
 669 |        "      <td>S</td>\n",
 670 |        "    </tr>\n",
 671 |        "    <tr>\n",
 672 |        "      <td>3</td>\n",
 673 |        "      <td>4</td>\n",
 674 |        "      <td>1</td>\n",
 675 |        "      <td>1</td>\n",
 676 |        "      <td>Futrelle, Mrs. Jacques Heath (Lily May Peel)</td>\n",
 677 |        "      <td>female</td>\n",
 678 |        "      <td>35.0</td>\n",
 679 |        "      <td>1</td>\n",
 680 |        "      <td>0</td>\n",
 681 |        "      <td>113803</td>\n",
 682 |        "      <td>53.1000</td>\n",
 683 |        "      <td>C123</td>\n",
 684 |        "      <td>S</td>\n",
 685 |        "    </tr>\n",
 686 |        "    <tr>\n",
 687 |        "      <td>4</td>\n",
 688 |        "      <td>5</td>\n",
 689 |        "      <td>0</td>\n",
 690 |        "      <td>3</td>\n",
 691 |        "      <td>Allen, Mr. William Henry</td>\n",
 692 |        "      <td>male</td>\n",
 693 |        "      <td>35.0</td>\n",
 694 |        "      <td>0</td>\n",
 695 |        "      <td>0</td>\n",
 696 |        "      <td>373450</td>\n",
 697 |        "      <td>8.0500</td>\n",
 698 |        "      <td>NaN</td>\n",
 699 |        "      <td>S</td>\n",
 700 |        "    </tr>\n",
 701 |        "  </tbody>\n",
 702 |        "</table>\n",
 703 |        "</div>"
 704 |       ],
 705 |       "text/plain": [
 706 |        "   PassengerId  Survived  Pclass  \\\n",
 707 |        "0            1         0       3   \n",
 708 |        "1            2         1       1   \n",
 709 |        "2            3         1       3   \n",
 710 |        "3            4         1       1   \n",
 711 |        "4            5         0       3   \n",
 712 |        "\n",
 713 |        "                                                Name     Sex   Age  SibSp  \\\n",
 714 |        "0                            Braund, Mr. Owen Harris    male  22.0      1   \n",
 715 |        "1  Cumings, Mrs. John Bradley (Florence Briggs Th...  female  38.0      1   \n",
 716 |        "2                             Heikkinen, Miss. Laina  female  26.0      0   \n",
 717 |        "3       Futrelle, Mrs. Jacques Heath (Lily May Peel)  female  35.0      1   \n",
 718 |        "4                           Allen, Mr. William Henry    male  35.0      0   \n",
 719 |        "\n",
 720 |        "   Parch            Ticket     Fare Cabin Embarked  \n",
 721 |        "0      0         A/5 21171   7.2500   NaN        S  \n",
 722 |        "1      0          PC 17599  71.2833   C85        C  \n",
 723 |        "2      0  STON/O2. 3101282   7.9250   NaN        S  \n",
 724 |        "3      0            113803  53.1000  C123        S  \n",
 725 |        "4      0            373450   8.0500   NaN        S  "
 726 |       ]
 727 |      },
 728 |      "execution_count": 18,
 729 |      "metadata": {},
 730 |      "output_type": "execute_result"
 731 |     }
 732 |    ],
 733 |    "source": [
 734 |     "# Датасет Титаника\n",
 735 |     "\n",
 736 |     "df = pd.read_csv('https://stepic.org/media/attachments/course/524/train.csv')\n",
 737 |     "df = df[df.Age.notnull()]\n",
 738 |     "df.head()"
 739 |    ]
 740 |   },
 741 |   {
 742 |    "cell_type": "code",
 743 |    "execution_count": 19,
 744 |    "metadata": {},
 745 |    "outputs": [
 746 |     {
 747 |      "data": {
 748 |       "text/html": [
 749 |        "<table class=\"simpletable\">\n",
 750 |        "<caption>Generalized Linear Model Regression Results</caption>\n",
 751 |        "<tr>\n",
 752 |        "  <th>Dep. Variable:</th>       <td>Survived</td>     <th>  No. Observations:  </th>  <td>   714</td> \n",
 753 |        "</tr>\n",
 754 |        "<tr>\n",
 755 |        "  <th>Model:</th>                  <td>GLM</td>       <th>  Df Residuals:      </th>  <td>   713</td> \n",
 756 |        "</tr>\n",
 757 |        "<tr>\n",
 758 |        "  <th>Model Family:</th>        <td>Binomial</td>     <th>  Df Model:          </th>  <td>     0</td> \n",
 759 |        "</tr>\n",
 760 |        "<tr>\n",
 761 |        "  <th>Link Function:</th>         <td>logit</td>      <th>  Scale:             </th> <td>  1.0000</td>\n",
 762 |        "</tr>\n",
 763 |        "<tr>\n",
 764 |        "  <th>Method:</th>                <td>IRLS</td>       <th>  Log-Likelihood:    </th> <td> -482.26</td>\n",
 765 |        "</tr>\n",
 766 |        "<tr>\n",
 767 |        "  <th>Date:</th>            <td>Tue, 08 Jun 2021</td> <th>  Deviance:          </th> <td>  964.52</td>\n",
 768 |        "</tr>\n",
 769 |        "<tr>\n",
 770 |        "  <th>Time:</th>                <td>20:53:33</td>     <th>  Pearson chi2:      </th>  <td>  714.</td> \n",
 771 |        "</tr>\n",
 772 |        "<tr>\n",
 773 |        "  <th>No. Iterations:</th>          <td>4</td>        <th>                     </th>     <td> </td>   \n",
 774 |        "</tr>\n",
 775 |        "<tr>\n",
 776 |        "  <th>Covariance Type:</th>     <td>nonrobust</td>    <th>                     </th>     <td> </td>   \n",
 777 |        "</tr>\n",
 778 |        "</table>\n",
 779 |        "<table class=\"simpletable\">\n",
 780 |        "<tr>\n",
 781 |        "      <td></td>         <th>coef</th>     <th>std err</th>      <th>z</th>      <th>P>|z|</th>  <th>[0.025</th>    <th>0.975]</th>  \n",
 782 |        "</tr>\n",
 783 |        "<tr>\n",
 784 |        "  <th>Intercept</th> <td>   -0.3799</td> <td>    0.076</td> <td>   -4.985</td> <td> 0.000</td> <td>   -0.529</td> <td>   -0.230</td>\n",
 785 |        "</tr>\n",
 786 |        "</table>"
 787 |       ],
 788 |       "text/plain": [
 789 |        "<class 'statsmodels.iolib.summary.Summary'>\n",
 790 |        "\"\"\"\n",
 791 |        "                 Generalized Linear Model Regression Results                  \n",
 792 |        "==============================================================================\n",
 793 |        "Dep. Variable:               Survived   No. Observations:                  714\n",
 794 |        "Model:                            GLM   Df Residuals:                      713\n",
 795 |        "Model Family:                Binomial   Df Model:                            0\n",
 796 |        "Link Function:                  logit   Scale:                          1.0000\n",
 797 |        "Method:                          IRLS   Log-Likelihood:                -482.26\n",
 798 |        "Date:                Tue, 08 Jun 2021   Deviance:                       964.52\n",
 799 |        "Time:                        20:53:33   Pearson chi2:                     714.\n",
 800 |        "No. Iterations:                     4                                         \n",
 801 |        "Covariance Type:            nonrobust                                         \n",
 802 |        "==============================================================================\n",
 803 |        "                 coef    std err          z      P>|z|      [0.025      0.975]\n",
 804 |        "------------------------------------------------------------------------------\n",
 805 |        "Intercept     -0.3799      0.076     -4.985      0.000      -0.529      -0.230\n",
 806 |        "==============================================================================\n",
 807 |        "\"\"\""
 808 |       ]
 809 |      },
 810 |      "execution_count": 19,
 811 |      "metadata": {},
 812 |      "output_type": "execute_result"
 813 |     }
 814 |    ],
 815 |    "source": [
 816 |     "glm_binomial = smf.glm(formula='Survived ~ 1', data=df, family=sm.families.Binomial()).fit()\n",
 817 |     "glm_binomial.summary()\n",
 818 |     "\n",
 819 |     "# Intercept - логарифм шансов (шансов выжить)"
 820 |    ]
 821 |   },
 822 |   {
 823 |    "cell_type": "code",
 824 |    "execution_count": 20,
 825 |    "metadata": {},
 826 |    "outputs": [],
 827 |    "source": [
 828 |     "# Получаем модель:\n",
 829 |     "# log(odds) = -0.3799"
 830 |    ]
 831 |   },
 832 |   {
 833 |    "cell_type": "code",
 834 |    "execution_count": 21,
 835 |    "metadata": {},
 836 |    "outputs": [
 837 |     {
 838 |      "data": {
 839 |       "text/plain": [
 840 |        "0    424\n",
 841 |        "1    290\n",
 842 |        "Name: Survived, dtype: int64"
 843 |       ]
 844 |      },
 845 |      "execution_count": 21,
 846 |      "metadata": {},
 847 |      "output_type": "execute_result"
 848 |     }
 849 |    ],
 850 |    "source": [
 851 |     "# Распределение частот ЗП\n",
 852 |     "df.Survived.value_counts()"
 853 |    ]
 854 |   },
 855 |   {
 856 |    "cell_type": "code",
 857 |    "execution_count": 22,
 858 |    "metadata": {},
 859 |    "outputs": [
 860 |     {
 861 |      "data": {
 862 |       "text/plain": [
 863 |        "<matplotlib.axes._subplots.AxesSubplot at 0x135ae516908>"
 864 |       ]
 865 |      },
 866 |      "execution_count": 22,
 867 |      "metadata": {},
 868 |      "output_type": "execute_result"
 869 |     },
 870 |     {
 871 |      "data": {
 872 |       "image/png": "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\n",
 873 |       "text/plain": [
 874 |        "<Figure size 432x288 with 1 Axes>"
 875 |       ]
 876 |      },
 877 |      "metadata": {
 878 |       "needs_background": "light"
 879 |      },
 880 |      "output_type": "display_data"
 881 |     }
 882 |    ],
 883 |    "source": [
 884 |     "sns.countplot(x='Survived', data=df, palette='hls')"
 885 |    ]
 886 |   },
 887 |   {
 888 |    "cell_type": "code",
 889 |    "execution_count": 23,
 890 |    "metadata": {},
 891 |    "outputs": [
 892 |     {
 893 |      "data": {
 894 |       "text/plain": [
 895 |        "0.6839622641509434"
 896 |       ]
 897 |      },
 898 |      "execution_count": 23,
 899 |      "metadata": {},
 900 |      "output_type": "execute_result"
 901 |     }
 902 |    ],
 903 |    "source": [
 904 |     "# df.Survived.value_counts()[1] - количество выживших 290\n",
 905 |     "# df.Survived.value_counts()[0] - количество погибших 424\n",
 906 |     "odds = df.Survived.value_counts()[1] / df.Survived.value_counts()[0]\n",
 907 |     "odds"
 908 |    ]
 909 |   },
 910 |   {
 911 |    "cell_type": "code",
 912 |    "execution_count": 24,
 913 |    "metadata": {},
 914 |    "outputs": [
 915 |     {
 916 |      "data": {
 917 |       "text/plain": [
 918 |        "-0.37985253225143806"
 919 |       ]
 920 |      },
 921 |      "execution_count": 24,
 922 |      "metadata": {},
 923 |      "output_type": "execute_result"
 924 |     }
 925 |    ],
 926 |    "source": [
 927 |     "# Логарифм шансов - это и есть Intercept.\n",
 928 |     "log_odds = np.log(odds)\n",
 929 |     "log_odds"
 930 |    ]
 931 |   },
 932 |   {
 933 |    "cell_type": "code",
 934 |    "execution_count": 25,
 935 |    "metadata": {},
 936 |    "outputs": [
 937 |     {
 938 |      "data": {
 939 |       "text/plain": [
 940 |        "0.6839622641509434"
 941 |       ]
 942 |      },
 943 |      "execution_count": 25,
 944 |      "metadata": {},
 945 |      "output_type": "execute_result"
 946 |     }
 947 |    ],
 948 |    "source": [
 949 |     "# Рассчитаем шансы:\n",
 950 |     "odds = np.exp(log_odds)\n",
 951 |     "odds"
 952 |    ]
 953 |   },
 954 |   {
 955 |    "cell_type": "code",
 956 |    "execution_count": 26,
 957 |    "metadata": {},
 958 |    "outputs": [
 959 |     {
 960 |      "data": {
 961 |       "text/plain": [
 962 |        "0.4061624649859944"
 963 |       ]
 964 |      },
 965 |      "execution_count": 26,
 966 |      "metadata": {},
 967 |      "output_type": "execute_result"
 968 |     }
 969 |    ],
 970 |    "source": [
 971 |     "# Рассчитаем вероятность:\n",
 972 |     "p = odds / (1 + odds)\n",
 973 |     "p"
 974 |    ]
 975 |   },
 976 |   {
 977 |    "cell_type": "code",
 978 |    "execution_count": 27,
 979 |    "metadata": {},
 980 |    "outputs": [
 981 |     {
 982 |      "data": {
 983 |       "text/plain": [
 984 |        "966.5159648555248"
 985 |       ]
 986 |      },
 987 |      "execution_count": 27,
 988 |      "metadata": {},
 989 |      "output_type": "execute_result"
 990 |     }
 991 |    ],
 992 |    "source": [
 993 |     "# aic - показатель остатков модели.\n",
 994 |     "# Чем он ниже, тем лучше модель (по аналогии с суммой квадратов остатков в линейной регрессии).\n",
 995 |     "glm_binomial.aic"
 996 |    ]
 997 |   },
 998 |   {
 999 |    "cell_type": "markdown",
1000 |    "metadata": {},
1001 |    "source": [
1002 |     "## Модель с одним номинативным предиктором"
1003 |    ]
1004 |   },
1005 |   {
1006 |    "cell_type": "code",
1007 |    "execution_count": 28,
1008 |    "metadata": {},
1009 |    "outputs": [
1010 |     {
1011 |      "data": {
1012 |       "text/html": [
1013 |        "<table class=\"simpletable\">\n",
1014 |        "<caption>Generalized Linear Model Regression Results</caption>\n",
1015 |        "<tr>\n",
1016 |        "  <th>Dep. Variable:</th>       <td>Survived</td>     <th>  No. Observations:  </th>  <td>   714</td> \n",
1017 |        "</tr>\n",
1018 |        "<tr>\n",
1019 |        "  <th>Model:</th>                  <td>GLM</td>       <th>  Df Residuals:      </th>  <td>   712</td> \n",
1020 |        "</tr>\n",
1021 |        "<tr>\n",
1022 |        "  <th>Model Family:</th>        <td>Binomial</td>     <th>  Df Model:          </th>  <td>     1</td> \n",
1023 |        "</tr>\n",
1024 |        "<tr>\n",
1025 |        "  <th>Link Function:</th>         <td>logit</td>      <th>  Scale:             </th> <td>  1.0000</td>\n",
1026 |        "</tr>\n",
1027 |        "<tr>\n",
1028 |        "  <th>Method:</th>                <td>IRLS</td>       <th>  Log-Likelihood:    </th> <td> -375.35</td>\n",
1029 |        "</tr>\n",
1030 |        "<tr>\n",
1031 |        "  <th>Date:</th>            <td>Tue, 08 Jun 2021</td> <th>  Deviance:          </th> <td>  750.70</td>\n",
1032 |        "</tr>\n",
1033 |        "<tr>\n",
1034 |        "  <th>Time:</th>                <td>20:53:34</td>     <th>  Pearson chi2:      </th>  <td>  714.</td> \n",
1035 |        "</tr>\n",
1036 |        "<tr>\n",
1037 |        "  <th>No. Iterations:</th>          <td>4</td>        <th>                     </th>     <td> </td>   \n",
1038 |        "</tr>\n",
1039 |        "<tr>\n",
1040 |        "  <th>Covariance Type:</th>     <td>nonrobust</td>    <th>                     </th>     <td> </td>   \n",
1041 |        "</tr>\n",
1042 |        "</table>\n",
1043 |        "<table class=\"simpletable\">\n",
1044 |        "<tr>\n",
1045 |        "         <td></td>           <th>coef</th>     <th>std err</th>      <th>z</th>      <th>P>|z|</th>  <th>[0.025</th>    <th>0.975]</th>  \n",
1046 |        "</tr>\n",
1047 |        "<tr>\n",
1048 |        "  <th>Intercept</th>      <td>    1.1243</td> <td>    0.144</td> <td>    7.814</td> <td> 0.000</td> <td>    0.842</td> <td>    1.406</td>\n",
1049 |        "</tr>\n",
1050 |        "<tr>\n",
1051 |        "  <th>C(Sex)[T.male]</th> <td>   -2.4778</td> <td>    0.185</td> <td>  -13.392</td> <td> 0.000</td> <td>   -2.840</td> <td>   -2.115</td>\n",
1052 |        "</tr>\n",
1053 |        "</table>"
1054 |       ],
1055 |       "text/plain": [
1056 |        "<class 'statsmodels.iolib.summary.Summary'>\n",
1057 |        "\"\"\"\n",
1058 |        "                 Generalized Linear Model Regression Results                  \n",
1059 |        "==============================================================================\n",
1060 |        "Dep. Variable:               Survived   No. Observations:                  714\n",
1061 |        "Model:                            GLM   Df Residuals:                      712\n",
1062 |        "Model Family:                Binomial   Df Model:                            1\n",
1063 |        "Link Function:                  logit   Scale:                          1.0000\n",
1064 |        "Method:                          IRLS   Log-Likelihood:                -375.35\n",
1065 |        "Date:                Tue, 08 Jun 2021   Deviance:                       750.70\n",
1066 |        "Time:                        20:53:34   Pearson chi2:                     714.\n",
1067 |        "No. Iterations:                     4                                         \n",
1068 |        "Covariance Type:            nonrobust                                         \n",
1069 |        "==================================================================================\n",
1070 |        "                     coef    std err          z      P>|z|      [0.025      0.975]\n",
1071 |        "----------------------------------------------------------------------------------\n",
1072 |        "Intercept          1.1243      0.144      7.814      0.000       0.842       1.406\n",
1073 |        "C(Sex)[T.male]    -2.4778      0.185    -13.392      0.000      -2.840      -2.115\n",
1074 |        "==================================================================================\n",
1075 |        "\"\"\""
1076 |       ]
1077 |      },
1078 |      "execution_count": 28,
1079 |      "metadata": {},
1080 |      "output_type": "execute_result"
1081 |     }
1082 |    ],
1083 |    "source": [
1084 |     "glm_binomial = smf.glm(formula='Survived ~ C(Sex)', data=df, family=sm.families.Binomial()).fit()\n",
1085 |     "glm_binomial.summary()\n",
1086 |     "\n",
1087 |     "# Intercept      - логарифм шансов для женщин\n",
1088 |     "# C(Sex)[T.male] - логарифм отношения шансов для мужчин к шансам для женщин"
1089 |    ]
1090 |   },
1091 |   {
1092 |    "cell_type": "code",
1093 |    "execution_count": 29,
1094 |    "metadata": {},
1095 |    "outputs": [
1096 |     {
1097 |      "data": {
1098 |       "text/html": [
1099 |        "<div>\n",
1100 |        "<style scoped>\n",
1101 |        "    .dataframe tbody tr th:only-of-type {\n",
1102 |        "        vertical-align: middle;\n",
1103 |        "    }\n",
1104 |        "\n",
1105 |        "    .dataframe tbody tr th {\n",
1106 |        "        vertical-align: top;\n",
1107 |        "    }\n",
1108 |        "\n",
1109 |        "    .dataframe thead th {\n",
1110 |        "        text-align: right;\n",
1111 |        "    }\n",
1112 |        "</style>\n",
1113 |        "<table border=\"1\" class=\"dataframe\">\n",
1114 |        "  <thead>\n",
1115 |        "    <tr style=\"text-align: right;\">\n",
1116 |        "      <th>Sex</th>\n",
1117 |        "      <th>female</th>\n",
1118 |        "      <th>male</th>\n",
1119 |        "    </tr>\n",
1120 |        "    <tr>\n",
1121 |        "      <th>Survived</th>\n",
1122 |        "      <th></th>\n",
1123 |        "      <th></th>\n",
1124 |        "    </tr>\n",
1125 |        "  </thead>\n",
1126 |        "  <tbody>\n",
1127 |        "    <tr>\n",
1128 |        "      <td>0</td>\n",
1129 |        "      <td>64</td>\n",
1130 |        "      <td>360</td>\n",
1131 |        "    </tr>\n",
1132 |        "    <tr>\n",
1133 |        "      <td>1</td>\n",
1134 |        "      <td>197</td>\n",
1135 |        "      <td>93</td>\n",
1136 |        "    </tr>\n",
1137 |        "  </tbody>\n",
1138 |        "</table>\n",
1139 |        "</div>"
1140 |       ],
1141 |       "text/plain": [
1142 |        "Sex       female  male\n",
1143 |        "Survived              \n",
1144 |        "0             64   360\n",
1145 |        "1            197    93"
1146 |       ]
1147 |      },
1148 |      "execution_count": 29,
1149 |      "metadata": {},
1150 |      "output_type": "execute_result"
1151 |     }
1152 |    ],
1153 |    "source": [
1154 |     "# Таблица сопряженности между переменными выжил/не выжил и пол\n",
1155 |     "cross_table = pd.crosstab(df.Survived, df.Sex)\n",
1156 |     "cross_table"
1157 |    ]
1158 |   },
1159 |   {
1160 |    "cell_type": "code",
1161 |    "execution_count": 30,
1162 |    "metadata": {},
1163 |    "outputs": [
1164 |     {
1165 |      "data": {
1166 |       "text/plain": [
1167 |        "Text(0, 0.5, 'Количество')"
1168 |       ]
1169 |      },
1170 |      "execution_count": 30,
1171 |      "metadata": {},
1172 |      "output_type": "execute_result"
1173 |     },
1174 |     {
1175 |      "data": {
1176 |       "image/png": "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\n",
1177 |       "text/plain": [
1178 |        "<Figure size 432x288 with 1 Axes>"
1179 |       ]
1180 |      },
1181 |      "metadata": {
1182 |       "needs_background": "light"
1183 |      },
1184 |      "output_type": "display_data"
1185 |     }
1186 |    ],
1187 |    "source": [
1188 |     "# Визуализировать таблицу сопряженности можно так:\n",
1189 |     "pd.crosstab(df.Sex, df.Survived).plot(kind='bar')\n",
1190 |     "plt.title('Количество мужчин и женщин на Титанике')\n",
1191 |     "plt.xlabel('Пол')\n",
1192 |     "plt.ylabel('Количество')"
1193 |    ]
1194 |   },
1195 |   {
1196 |    "cell_type": "code",
1197 |    "execution_count": 31,
1198 |    "metadata": {},
1199 |    "outputs": [
1200 |     {
1201 |      "data": {
1202 |       "text/plain": [
1203 |        "3.078125"
1204 |       ]
1205 |      },
1206 |      "execution_count": 31,
1207 |      "metadata": {},
1208 |      "output_type": "execute_result"
1209 |     }
1210 |    ],
1211 |    "source": [
1212 |     "# cross_table.female[0] - количество погибших женщин\n",
1213 |     "# cross_table.female[1] - количество выживших женщин\n",
1214 |     "odds_female = cross_table.female[1] / cross_table.female[0]\n",
1215 |     "odds_female"
1216 |    ]
1217 |   },
1218 |   {
1219 |    "cell_type": "code",
1220 |    "execution_count": 32,
1221 |    "metadata": {},
1222 |    "outputs": [
1223 |     {
1224 |      "data": {
1225 |       "text/plain": [
1226 |        "1.1243206453783166"
1227 |       ]
1228 |      },
1229 |      "execution_count": 32,
1230 |      "metadata": {},
1231 |      "output_type": "execute_result"
1232 |     }
1233 |    ],
1234 |    "source": [
1235 |     "# Логарифм шансов для женщин - это и есть Intercept\n",
1236 |     "np.log(odds_female)"
1237 |    ]
1238 |   },
1239 |   {
1240 |    "cell_type": "code",
1241 |    "execution_count": 33,
1242 |    "metadata": {},
1243 |    "outputs": [
1244 |     {
1245 |      "data": {
1246 |       "text/plain": [
1247 |        "0.25833333333333336"
1248 |       ]
1249 |      },
1250 |      "execution_count": 33,
1251 |      "metadata": {},
1252 |      "output_type": "execute_result"
1253 |     }
1254 |    ],
1255 |    "source": [
1256 |     "# cross_table.male[0] - количество погибших мужчин\n",
1257 |     "# cross_table.male[1] - количество выживших мужчин\n",
1258 |     "odds_male = cross_table.male[1] / cross_table.male[0]\n",
1259 |     "odds_male"
1260 |    ]
1261 |   },
1262 |   {
1263 |    "cell_type": "code",
1264 |    "execution_count": 34,
1265 |    "metadata": {},
1266 |    "outputs": [
1267 |     {
1268 |      "data": {
1269 |       "text/plain": [
1270 |        "0.08392554991539763"
1271 |       ]
1272 |      },
1273 |      "execution_count": 34,
1274 |      "metadata": {},
1275 |      "output_type": "execute_result"
1276 |     }
1277 |    ],
1278 |    "source": [
1279 |     "odds_ratio = odds_male / odds_female\n",
1280 |     "odds_ratio"
1281 |    ]
1282 |   },
1283 |   {
1284 |    "cell_type": "code",
1285 |    "execution_count": 35,
1286 |    "metadata": {},
1287 |    "outputs": [
1288 |     {
1289 |      "data": {
1290 |       "text/plain": [
1291 |        "-2.4778251836752165"
1292 |       ]
1293 |      },
1294 |      "execution_count": 35,
1295 |      "metadata": {},
1296 |      "output_type": "execute_result"
1297 |     }
1298 |    ],
1299 |    "source": [
1300 |     "np.log(odds_ratio)\n",
1301 |     "# Логарифм отношения шансов для мужчин к шансам для женщин - это и есть значение C(Sex)[T.male]"
1302 |    ]
1303 |   },
1304 |   {
1305 |    "cell_type": "code",
1306 |    "execution_count": 36,
1307 |    "metadata": {},
1308 |    "outputs": [
1309 |     {
1310 |      "data": {
1311 |       "text/plain": [
1312 |        "-1.3534999999999997"
1313 |       ]
1314 |      },
1315 |      "execution_count": 36,
1316 |      "metadata": {},
1317 |      "output_type": "execute_result"
1318 |     }
1319 |    ],
1320 |    "source": [
1321 |     "# Получаем модель:\n",
1322 |     "# log(odds) = 1.1243 - (2.4778 * Sex_male)\n",
1323 |     "\n",
1324 |     "# Если предсказываем логарифм шансов для женщин (Sex_male = 0):\n",
1325 |     "# log(odds) = 1.1243 - (2.4778 * 0)\n",
1326 |     "# log(odds) = 1.1243\n",
1327 |     "\n",
1328 |     "# Если предсказываем логарифм шансов для мужчин (Sex_male = 1):\n",
1329 |     "# log(odds) = 1.1243 - (2.4778 * 1)\n",
1330 |     "# log(odds) = 1.1243 - 2.4778\n",
1331 |     "# log(odds) -1.3535\n",
1332 |     "\n",
1333 |     "log_odds_male = 1.1243 - 2.4778\n",
1334 |     "log_odds_male"
1335 |    ]
1336 |   },
1337 |   {
1338 |    "cell_type": "code",
1339 |    "execution_count": 37,
1340 |    "metadata": {},
1341 |    "outputs": [
1342 |     {
1343 |      "data": {
1344 |       "text/plain": [
1345 |        "754.7002061466517"
1346 |       ]
1347 |      },
1348 |      "execution_count": 37,
1349 |      "metadata": {},
1350 |      "output_type": "execute_result"
1351 |     }
1352 |    ],
1353 |    "source": [
1354 |     "# aic у этой модели ниже, чем у предыдущей, значит эта модель лучше.\n",
1355 |     "glm_binomial.aic"
1356 |    ]
1357 |   },
1358 |   {
1359 |    "cell_type": "markdown",
1360 |    "metadata": {},
1361 |    "source": [
1362 |     "## Модель с двумя номинативными предикторами"
1363 |    ]
1364 |   },
1365 |   {
1366 |    "cell_type": "code",
1367 |    "execution_count": 38,
1368 |    "metadata": {},
1369 |    "outputs": [
1370 |     {
1371 |      "data": {
1372 |       "text/html": [
1373 |        "<table class=\"simpletable\">\n",
1374 |        "<caption>Generalized Linear Model Regression Results</caption>\n",
1375 |        "<tr>\n",
1376 |        "  <th>Dep. Variable:</th>       <td>Survived</td>     <th>  No. Observations:  </th>  <td>   714</td> \n",
1377 |        "</tr>\n",
1378 |        "<tr>\n",
1379 |        "  <th>Model:</th>                  <td>GLM</td>       <th>  Df Residuals:      </th>  <td>   708</td> \n",
1380 |        "</tr>\n",
1381 |        "<tr>\n",
1382 |        "  <th>Model Family:</th>        <td>Binomial</td>     <th>  Df Model:          </th>  <td>     5</td> \n",
1383 |        "</tr>\n",
1384 |        "<tr>\n",
1385 |        "  <th>Link Function:</th>         <td>logit</td>      <th>  Scale:             </th> <td>  1.0000</td>\n",
1386 |        "</tr>\n",
1387 |        "<tr>\n",
1388 |        "  <th>Method:</th>                <td>IRLS</td>       <th>  Log-Likelihood:    </th> <td> -321.14</td>\n",
1389 |        "</tr>\n",
1390 |        "<tr>\n",
1391 |        "  <th>Date:</th>            <td>Tue, 08 Jun 2021</td> <th>  Deviance:          </th> <td>  642.28</td>\n",
1392 |        "</tr>\n",
1393 |        "<tr>\n",
1394 |        "  <th>Time:</th>                <td>20:53:34</td>     <th>  Pearson chi2:      </th>  <td>  714.</td> \n",
1395 |        "</tr>\n",
1396 |        "<tr>\n",
1397 |        "  <th>No. Iterations:</th>          <td>6</td>        <th>                     </th>     <td> </td>   \n",
1398 |        "</tr>\n",
1399 |        "<tr>\n",
1400 |        "  <th>Covariance Type:</th>     <td>nonrobust</td>    <th>                     </th>     <td> </td>   \n",
1401 |        "</tr>\n",
1402 |        "</table>\n",
1403 |        "<table class=\"simpletable\">\n",
1404 |        "<tr>\n",
1405 |        "                <td></td>                   <th>coef</th>     <th>std err</th>      <th>z</th>      <th>P>|z|</th>  <th>[0.025</th>    <th>0.975]</th>  \n",
1406 |        "</tr>\n",
1407 |        "<tr>\n",
1408 |        "  <th>Intercept</th>                     <td>    3.3081</td> <td>    0.588</td> <td>    5.628</td> <td> 0.000</td> <td>    2.156</td> <td>    4.460</td>\n",
1409 |        "</tr>\n",
1410 |        "<tr>\n",
1411 |        "  <th>C(Sex)[T.male]</th>                <td>   -3.7301</td> <td>    0.622</td> <td>   -5.997</td> <td> 0.000</td> <td>   -4.949</td> <td>   -2.511</td>\n",
1412 |        "</tr>\n",
1413 |        "<tr>\n",
1414 |        "  <th>C(Pclass)[T.2]</th>                <td>   -0.8804</td> <td>    0.726</td> <td>   -1.213</td> <td> 0.225</td> <td>   -2.303</td> <td>    0.542</td>\n",
1415 |        "</tr>\n",
1416 |        "<tr>\n",
1417 |        "  <th>C(Pclass)[T.3]</th>                <td>   -3.4653</td> <td>    0.620</td> <td>   -5.585</td> <td> 0.000</td> <td>   -4.681</td> <td>   -2.249</td>\n",
1418 |        "</tr>\n",
1419 |        "<tr>\n",
1420 |        "  <th>C(Sex)[T.male]:C(Pclass)[T.2]</th> <td>   -0.4204</td> <td>    0.804</td> <td>   -0.523</td> <td> 0.601</td> <td>   -1.997</td> <td>    1.156</td>\n",
1421 |        "</tr>\n",
1422 |        "<tr>\n",
1423 |        "  <th>C(Sex)[T.male]:C(Pclass)[T.3]</th> <td>    2.1542</td> <td>    0.676</td> <td>    3.185</td> <td> 0.001</td> <td>    0.829</td> <td>    3.480</td>\n",
1424 |        "</tr>\n",
1425 |        "</table>"
1426 |       ],
1427 |       "text/plain": [
1428 |        "<class 'statsmodels.iolib.summary.Summary'>\n",
1429 |        "\"\"\"\n",
1430 |        "                 Generalized Linear Model Regression Results                  \n",
1431 |        "==============================================================================\n",
1432 |        "Dep. Variable:               Survived   No. Observations:                  714\n",
1433 |        "Model:                            GLM   Df Residuals:                      708\n",
1434 |        "Model Family:                Binomial   Df Model:                            5\n",
1435 |        "Link Function:                  logit   Scale:                          1.0000\n",
1436 |        "Method:                          IRLS   Log-Likelihood:                -321.14\n",
1437 |        "Date:                Tue, 08 Jun 2021   Deviance:                       642.28\n",
1438 |        "Time:                        20:53:34   Pearson chi2:                     714.\n",
1439 |        "No. Iterations:                     6                                         \n",
1440 |        "Covariance Type:            nonrobust                                         \n",
1441 |        "=================================================================================================\n",
1442 |        "                                    coef    std err          z      P>|z|      [0.025      0.975]\n",
1443 |        "-------------------------------------------------------------------------------------------------\n",
1444 |        "Intercept                         3.3081      0.588      5.628      0.000       2.156       4.460\n",
1445 |        "C(Sex)[T.male]                   -3.7301      0.622     -5.997      0.000      -4.949      -2.511\n",
1446 |        "C(Pclass)[T.2]                   -0.8804      0.726     -1.213      0.225      -2.303       0.542\n",
1447 |        "C(Pclass)[T.3]                   -3.4653      0.620     -5.585      0.000      -4.681      -2.249\n",
1448 |        "C(Sex)[T.male]:C(Pclass)[T.2]    -0.4204      0.804     -0.523      0.601      -1.997       1.156\n",
1449 |        "C(Sex)[T.male]:C(Pclass)[T.3]     2.1542      0.676      3.185      0.001       0.829       3.480\n",
1450 |        "=================================================================================================\n",
1451 |        "\"\"\""
1452 |       ]
1453 |      },
1454 |      "execution_count": 38,
1455 |      "metadata": {},
1456 |      "output_type": "execute_result"
1457 |     }
1458 |    ],
1459 |    "source": [
1460 |     "glm_binomial = smf.glm(formula='Survived ~ C(Sex) * C(Pclass)', data=df, family=sm.families.Binomial()).fit()\n",
1461 |     "glm_binomial.summary()\n",
1462 |     "\n",
1463 |     "# Intercept      - логарифм шансов для Ж в 1кл\n",
1464 |     "# C(Sex)[T.male] - логарифм отношения шансов для М в 1кл к шансам Ж в 1кл\n",
1465 |     "# C(Pclass)[T.2] - логарифм отношения шансов для Ж во 2кл к шансам Ж в 1кл\n",
1466 |     "# C(Pclass)[T.3] - логарифм отношения шансов для Ж в 3кл к шансам Ж в 1кл\n",
1467 |     "# C(Sex)[T.male]:C(Pclass)[T.2] - разность логарифмов отношения шансов, рассчитанных для М/Ж во 2кл и М/Ж в 1кл\n",
1468 |     "# C(Sex)[T.male]:C(Pclass)[T.3] - разность логарифмов отношения шансов, рассчитанных для М/Ж во 3кл и М/Ж в 1кл"
1469 |    ]
1470 |   },
1471 |   {
1472 |    "cell_type": "code",
1473 |    "execution_count": 39,
1474 |    "metadata": {},
1475 |    "outputs": [
1476 |     {
1477 |      "data": {
1478 |       "text/html": [
1479 |        "<div>\n",
1480 |        "<style scoped>\n",
1481 |        "    .dataframe tbody tr th:only-of-type {\n",
1482 |        "        vertical-align: middle;\n",
1483 |        "    }\n",
1484 |        "\n",
1485 |        "    .dataframe tbody tr th {\n",
1486 |        "        vertical-align: top;\n",
1487 |        "    }\n",
1488 |        "\n",
1489 |        "    .dataframe thead th {\n",
1490 |        "        text-align: right;\n",
1491 |        "    }\n",
1492 |        "</style>\n",
1493 |        "<table border=\"1\" class=\"dataframe\">\n",
1494 |        "  <thead>\n",
1495 |        "    <tr style=\"text-align: right;\">\n",
1496 |        "      <th>Pclass</th>\n",
1497 |        "      <th>1</th>\n",
1498 |        "      <th>2</th>\n",
1499 |        "      <th>3</th>\n",
1500 |        "    </tr>\n",
1501 |        "    <tr>\n",
1502 |        "      <th>Survived</th>\n",
1503 |        "      <th></th>\n",
1504 |        "      <th></th>\n",
1505 |        "      <th></th>\n",
1506 |        "    </tr>\n",
1507 |        "  </thead>\n",
1508 |        "  <tbody>\n",
1509 |        "    <tr>\n",
1510 |        "      <td>0</td>\n",
1511 |        "      <td>3</td>\n",
1512 |        "      <td>6</td>\n",
1513 |        "      <td>55</td>\n",
1514 |        "    </tr>\n",
1515 |        "    <tr>\n",
1516 |        "      <td>1</td>\n",
1517 |        "      <td>82</td>\n",
1518 |        "      <td>68</td>\n",
1519 |        "      <td>47</td>\n",
1520 |        "    </tr>\n",
1521 |        "  </tbody>\n",
1522 |        "</table>\n",
1523 |        "</div>"
1524 |       ],
1525 |       "text/plain": [
1526 |        "Pclass     1   2   3\n",
1527 |        "Survived            \n",
1528 |        "0          3   6  55\n",
1529 |        "1         82  68  47"
1530 |       ]
1531 |      },
1532 |      "execution_count": 39,
1533 |      "metadata": {},
1534 |      "output_type": "execute_result"
1535 |     }
1536 |    ],
1537 |    "source": [
1538 |     "# Таблица сопряженности между переменными выжил/не выжил и классом пассажира для женщин.\n",
1539 |     "cross_table_female = pd.crosstab(df[df.Sex == 'female'].Survived, df.Pclass)\n",
1540 |     "cross_table_female"
1541 |    ]
1542 |   },
1543 |   {
1544 |    "cell_type": "code",
1545 |    "execution_count": 40,
1546 |    "metadata": {},
1547 |    "outputs": [
1548 |     {
1549 |      "data": {
1550 |       "text/plain": [
1551 |        "Text(0, 0.5, 'Количество')"
1552 |       ]
1553 |      },
1554 |      "execution_count": 40,
1555 |      "metadata": {},
1556 |      "output_type": "execute_result"
1557 |     },
1558 |     {
1559 |      "data": {
1560 |       "image/png": "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\n",
1561 |       "text/plain": [
1562 |        "<Figure size 432x288 with 1 Axes>"
1563 |       ]
1564 |      },
1565 |      "metadata": {
1566 |       "needs_background": "light"
1567 |      },
1568 |      "output_type": "display_data"
1569 |     }
1570 |    ],
1571 |    "source": [
1572 |     "pd.crosstab(df.Pclass[df.Sex == 'female'], df.Survived).plot(kind='bar')\n",
1573 |     "plt.title('Выживаемость женщин в разных классах')\n",
1574 |     "plt.xlabel('Класс пассажира')\n",
1575 |     "plt.ylabel('Количество')"
1576 |    ]
1577 |   },
1578 |   {
1579 |    "cell_type": "code",
1580 |    "execution_count": 41,
1581 |    "metadata": {},
1582 |    "outputs": [
1583 |     {
1584 |      "data": {
1585 |       "text/html": [
1586 |        "<div>\n",
1587 |        "<style scoped>\n",
1588 |        "    .dataframe tbody tr th:only-of-type {\n",
1589 |        "        vertical-align: middle;\n",
1590 |        "    }\n",
1591 |        "\n",
1592 |        "    .dataframe tbody tr th {\n",
1593 |        "        vertical-align: top;\n",
1594 |        "    }\n",
1595 |        "\n",
1596 |        "    .dataframe thead th {\n",
1597 |        "        text-align: right;\n",
1598 |        "    }\n",
1599 |        "</style>\n",
1600 |        "<table border=\"1\" class=\"dataframe\">\n",
1601 |        "  <thead>\n",
1602 |        "    <tr style=\"text-align: right;\">\n",
1603 |        "      <th>Pclass</th>\n",
1604 |        "      <th>1</th>\n",
1605 |        "      <th>2</th>\n",
1606 |        "      <th>3</th>\n",
1607 |        "    </tr>\n",
1608 |        "    <tr>\n",
1609 |        "      <th>Survived</th>\n",
1610 |        "      <th></th>\n",
1611 |        "      <th></th>\n",
1612 |        "      <th></th>\n",
1613 |        "    </tr>\n",
1614 |        "  </thead>\n",
1615 |        "  <tbody>\n",
1616 |        "    <tr>\n",
1617 |        "      <td>0</td>\n",
1618 |        "      <td>61</td>\n",
1619 |        "      <td>84</td>\n",
1620 |        "      <td>215</td>\n",
1621 |        "    </tr>\n",
1622 |        "    <tr>\n",
1623 |        "      <td>1</td>\n",
1624 |        "      <td>40</td>\n",
1625 |        "      <td>15</td>\n",
1626 |        "      <td>38</td>\n",
1627 |        "    </tr>\n",
1628 |        "  </tbody>\n",
1629 |        "</table>\n",
1630 |        "</div>"
1631 |       ],
1632 |       "text/plain": [
1633 |        "Pclass     1   2    3\n",
1634 |        "Survived             \n",
1635 |        "0         61  84  215\n",
1636 |        "1         40  15   38"
1637 |       ]
1638 |      },
1639 |      "execution_count": 41,
1640 |      "metadata": {},
1641 |      "output_type": "execute_result"
1642 |     }
1643 |    ],
1644 |    "source": [
1645 |     "# Таблица сопряженности между переменными выжил/не выжил и классом пассажира для мужчин.\n",
1646 |     "cross_table_male = pd.crosstab(df[df.Sex == 'male'].Survived, df.Pclass)\n",
1647 |     "cross_table_male"
1648 |    ]
1649 |   },
1650 |   {
1651 |    "cell_type": "code",
1652 |    "execution_count": 42,
1653 |    "metadata": {},
1654 |    "outputs": [
1655 |     {
1656 |      "data": {
1657 |       "text/plain": [
1658 |        "Text(0, 0.5, 'Количество')"
1659 |       ]
1660 |      },
1661 |      "execution_count": 42,
1662 |      "metadata": {},
1663 |      "output_type": "execute_result"
1664 |     },
1665 |     {
1666 |      "data": {
1667 |       "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAETCAYAAADZHBoWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nO3debxVdb3/8ddbQDBRlMmEAx5K8yeoUIJDDlF6HVDRW8rQAGZeTCW5zdbNK91flpnerlPdSA26DohpgcPPMm9a5pAexVkSk+QAqYAiqIjA5/fH+p7l9rDPOfvo2Xuf4f18PM7jrPVd02ettff+rO93TYoIzMzMALaqdgBmZtZ+OCmYmVnOScHMzHJOCmZmlnNSMDOznJOCmZnlnBTMzCznpNCGJC2R9IakdZJelnSLpCHVjsvMrFROCm3v2IjoDewMvABcUuV4zMxK5qRQJhGxHvgVMLyhTNJsSd9rPK6kUyTdmbr7S1ok6YTUP1ZSfereStINkn6c+mslhaTuqX/f1P+9gmk3p5rLWkl/kbRnwXKvl/QPSWsk/VHSiIJhPSVdIOl5SS9I+m9J2xTMNyRdVDD+8FR2VUHZeElPSHpF0p2S9igYNkTSjZJekrRK0qWSBqVY10naIOmtgv6DC7dFS0qJMdXkvtRoukclHd/K/XCSpLsL5tG4PyTtmrqHptpkQwzv2Iep7CpJM5tYr5MkbUrb5FVJ/ytpcDPj/lnSJWkfPy3p0ILhn5f0VPps/E3SqQXDDpT0XFrOUklnFAy7U9IpBf2HSVpS0L9E0mFF4snXS9I3Jd1X8Nk9LX1WehWZ7h37XdL5ku4qHDfFtD7Fu77R9m/uc76NpAsl/T0Nv7vgc36QpHvS53eppJNS+dGSHk7bf2nhvpI0MW3L7VP/UWnZA4rto/bISaFMJL0PmAjc15rpImIlcDTwn5I+2mjwhYCArzYx+fnAskZly1PNZQfgEWBmwbD/B+wGDAQeAq4uGPZD4EPAKGBXYDDw7wXDXwKOktQz9Z8CPNUwUNKHgGuBfwUGALcCN0naWlI34Gbg70BtmvfciFgeEb1TvN8Hrmvoj4g/NbHOzWk2RmAO8NmCmEemWG5t5X7YTOnfpf8LrGrNShRxb9pGA4E3gS83M+5+wN+A/sA5wI2S+qZhLwLHANsDnwd+LOkjadhfgYPTcsYDP5LU5z3GXehHwAbgO5J2I9vfn00HU02S9E3gMLIaeeG4WwFnpHi/2Giy5j7nFwD7AB8F+gLfADZLGpqmu4Ts8zsKWJimeQ2YQvadOho4TdLxABFxHXAvcLGkfsAVwCkR8VIpG6U9cFJoe7+R9ArwKvBPZB/+VomIxcBUYAHZh5l0RHs08JmI2Nx4GknHkO3P3zcx262AbhT8IEXElRGxNiLeJEsWIyX1kSTgX4AvR8TqiFhL9qWdVDC/DWQ/9J+UtDVwFPCbguETgVsi4vaIeIvsy7cN2ZdvX2AQ8PWIeC0i1kfE3bS9lmKcD+yWfpQAPkeWiDZAq/bD88AekmqaC0bS3sABZMmoLWyV/ppLMi8C/xURb6UfrEVk8RMRt0TEs5G5C/gdcHAa9lJENBydK033WhvFTdp2U4Azybbv+RHxcHPTpNrJ14AjI+LVRoO3JtvfxZbV1Od8K+BkYEZELIuITRFxTxrvM8DvI+LatO1WRcTCNL87I+KxiNgcEY+SHfx8rGCRZwCfAO4EboqIm0vfMtXnpND2jo+IHYCewHTgLknvLxj+tVQdfUFZ80m/JubzT8AS4L/IjlTOBLYjO2pvbCvgB2RHOY0NSklqLdmP4iUAkrpJOk/Ss5JeTcuC7IhyAPA+oC7F+gpwWyovdDnwBeD4NLzwSzmIrCYA5D8CS8mOxIcAf4+IjU2se3MGpZheTlX4I1oYv8kY05d/HvDZ9AMxGfifRtOXsh/uAuYCj6Rt9ZMmYvkhcDbwVpFhKwu29YQW1mn/NN4rwDBgdjPjLot3PvXy72T7pqFp4z5Jq9P8xpHtf9LwgyStBe4HftNof11cEG9hom3wm4Jml/PTgcY7RMQS4A9ktcXLWljnAWTb7nWyo/bG+gIvNy5s4XPeH+gFPFtkfkOaKEfSfpL+oKzpcw1ZzSTfbhHxCnA9sCdZrbJDcVIok3TUcSOwCTioYNAFKWl8gOyH9+uNp5U0nKw6fxRwbprHMcC3gJ8V+YKdBCyKiGJNVcvT8rYBzgJuSOWfBo4jq4r3IftiQnZUuBJ4AxgRETukvz6pal64jo+ndfgO2Y/vO5YL7FKwTiL7oi0jSw5DVdCO3goN69OXLME1e9TdQoyk6T8DHAq8HhH3FsRc0n5IR9qnRUS/FNvpRZbzCbIfjnlNhNq/YVs3M06D+9J4vYCraD4pDG70eRkKLE9NajeQ1eB2SvO7lWz/k9br7ojYjuy82GmSjiqYz5kF8R5fZLkNB0cfJWui2yJ5SxpHVnO6g5Zr1JvI9sM0YJak7QrmszXZZ+2vRaZr6XO+HvhgkemWNlEOcA1Z7WZIRPQB/puC7SZpFFkN5Frg4hbWq91xUigTZY4DduSd7dgN1pMd9WzVeDrgZ8B3UjvkPcDqiFgUEbPJ2pBPbTSvfyP7oWpSOlrczNtHNNulea0i+9H8fsG4m4Gfk7UxD0xxDW7iqPz7ZNXsJxqVzwOOlnSopB5k7e9vpvX5C7ACOE/StpJ6STqwufibWJ9XKO0z3FSMpCSwmeyILq8lvMv90JyZZM1lbfms+iD7sWzuJOZA4ExJPSSdCOxB9uO/NVlt9iVgY/rBP7xhIkkfkNRwENCTbDu/8S5iXAtsZMvPeX9SeztZE92xKUk0ZXVEPBkRvyVLIuen+fQiO9e1OCKKJYWWPudXkp03GpRqFQekhHk1cJikCZK6S+qXfuwb5rk6ItZL2pcs8TSsV0Oi/jbZAcVgScUOEtotJ4W2d5OkdWTnFM4Fpjb6MTpT2ZUUz5Md6V3QaPovkH3Zr2xi/l8EZjZqkro5Ip5pYvyGK3rWkn1QT07lvyRrSlgGPMmWJ8S/CSwG7kvV7t8DuzeeeUTcHBFfKVK+iOwI8RKyI7JjyU4OboiITal/V7LtUE92DqIU75dUn7bh98i2V7OairHAL4G9yL7MDd7NfmjOwxFxZ4njtuSA9BlbA3ySrJmyKfeTnQ9ZSfZ5PCG1j68lawqbR9bs8mmyo98GY4G/puXcDPy4lfFfm/bR42TNS7c1Gj4LmB8Rt0bEKrLtfXkzzamFvgIcI2ksWQ3wo8AJTYzb0uf8a8BjwAPAarImvq0i4nmy5rSvpvKFwMg0zenAf6Tv1L/zzprdD4D6iPhpap78LPC9gvNW7Z7a9sDFrOORNAWYFhEHtThyB6LsEspTOtt6WXm5pmBdmrJLh08nO3I16/KcFKzLSudIXiK78/yaKodj1i64+cjMzHKuKZiZWc5JwczMcu/m5qF2o3///lFbW1vtMMzMOpS6urqVEVH0/pYOnRRqa2t58MEHqx2GmVmHIunvTQ1z85GZmeWcFMzMLOekYGZmuQ59TqGYt956i/r6etavb/ZdHR1Cr169qKmpoUePHtUOxcy6iE6XFOrr69luu+2ora2lyCPcO4yIYNWqVdTX1zNs2LBqh2NmXUSnaz5av349/fr169AJAUAS/fr16xQ1HjPrODpdUgA6fEJo0FnWw8w6jk6ZFIo599xzGTFiBHvvvTejRo3i/vvvf8/zXLBgAeedd14bRAe9e/dueSQzszLrdOcUirn33nu5+eabeeihh+jZsycrV65kw4ai7/jewsaNG+nevfhmGj9+POPHj2/LUM2siNqzbqno8pacd3RFl9eedImawooVK+jfvz89e/YEoH///gwaNIja2lpWrlwJwIMPPsjYsWMBmDlzJtOmTePwww9nypQp7LfffjzxxNsvTxs7dix1dXXMnj2b6dOns2bNGmpra9m8eTMAr7/+OkOGDOGtt97i2Wef5cgjj2Sfffbh4IMP5umnnwbgueee44ADDmDMmDGcffbZFdwaZmZN6xJJ4fDDD2fp0qV86EMf4vTTT+euu+5qcZq6ujrmz5/PNddcw6RJk5g3L3vj3ooVK1i+fDn77LNPPm6fPn0YOXJkPt+bbrqJI444gh49ejBt2jQuueQS6urquOCCCzj99Ox1rTNmzOC0007jgQce4P3vL/WNjmZm5dUlkkLv3r2pq6tj1qxZDBgwgIkTJzJ79uxmpxk/fjzbbLMNABMmTOD6668HYN68eZx44olbjD9x4kSuu+46AObOncvEiRNZt24d99xzDyeeeCKjRo3i1FNPZcWKFQD8+c9/ZvLkyQB87nOfa6tVNTN7T7rEOQWAbt26MXbsWMaOHctee+3FnDlz6N69e97k0/jSz2233TbvHjx4MP369ePRRx/luuuu42c/+9kW8x8/fjzf+ta3WL16NXV1dXziE5/gtddeY4cddmDhwoVFY/LVRWbW3nSJmsKiRYt45pln8v6FCxeyyy67UFtbS11dHQA33HBDs/OYNGkS559/PmvWrGGvvfbaYnjv3r3Zd999mTFjBscccwzdunVj++23Z9iwYXktIyJ45JFHADjwwAOZO3cuAFdffXWbrKeZ2XvVJZLCunXrmDp1KsOHD2fvvffmySefZObMmZxzzjnMmDGDgw8+mG7dujU7jxNOOIG5c+cyYcKEJseZOHEiV111FRMnTszLrr76aq644gpGjhzJiBEjmD9/PgAXXXQRl112GWPGjGHNmjVts6JmZu9Rh35H8+jRo6Px+xSeeuop9thjjypF1PY62/qYvRu+JLVtSaqLiNHFhnWJmoKZmZXGScHMzHJOCmZmlitbUpA0RNIfJD0l6QlJM1J5X0m3S3om/d8xlUvSxZIWS3pU0kfKFZuZmRVXzprCRuCrEbEHsD9whqThwFnAHRGxG3BH6gc4Ctgt/U0DflrG2MzMrIiyJYWIWBERD6XutcBTwGDgOGBOGm0OcHzqPg74ZWTuA3aQtHO54jMzsy1V5JyCpFrgw8D9wE4RsQKyxAEMTKMNBpYWTFafyjqN2267jd13351dd921zR65bWbWlsr+mAtJvYEbgH+NiFebebRDsQFb3EQhaRpZ8xJDhw59VzG19TXPpVzTvGnTJs444wxuv/12ampqGDNmDOPHj2f48OFtGouZ2XtR1pqCpB5kCeHqiLgxFb/Q0CyU/r+YyuuBIQWT1wDLG88zImZFxOiIGD1gwIDyBd/G/vKXv7DrrrvygQ98gK233ppJkybldzebmbUX5bz6SMAVwFMR8Z8FgxYAU1P3VGB+QfmUdBXS/sCahmamzmDZsmUMGfJ2zqupqWHZsmVVjMjMbEvlbD46EPgc8JikhseEfhs4D5gn6QvA80DDc6hvBcYBi4HXgc+XMbaKK/Y4ET8l1czam7IlhYi4m+LnCQAOLTJ+AGeUK55qq6mpYenSt8+j19fXM2jQoCpGZGa2Jd/RXCFjxozhmWee4bnnnmPDhg3MnTvX73c2s3any7xkp9q6d+/OpZdeyhFHHMGmTZs4+eSTGTFiRLXDMjN7hy6ZFKr1WNxx48Yxbty4qizbzKwUbj4yM7Ock4KZmeWcFMzMLOekYGZmOScFMzPLOSmYmVnOSaFCTj75ZAYOHMiee+5Z7VDMzJrUJe9TYGafNp7fmhZHOemkk5g+fTpTpkxp22WbmbUh1xQq5JBDDqFv377VDsPMrFlOCmZmlnNSMDOznJOCmZnlnBTMzCznpFAhkydP5oADDmDRokXU1NRwxRVXVDskM7MtdNFLUlu+hLStXXvttRVfpplZa7mmYGZmOScFMzPLOSmYmVmuUyaFiKh2CG2is6yHmXUcnS4p9OrVi1WrVnX4H9SIYNWqVfTq1avaoZhZF9Lprj6qqamhvr6el156qdqhvGe9evWipqam2mGYWRfS6ZJCjx49GDZsWLXDMDPrkDpd85GZmb17TgpmZpZzUjAzs5yTgpmZ5ZwUzMws56RgZmY5JwUzM8s5KZiZWc5JwczMck4KZmaWc1IwM7Ock4KZmeWcFMzMLOekYGZmubIlBUlXSnpR0uMFZTMlLZO0MP2NKxj2LUmLJS2SdES54jIzs6aVs6YwGziySPmPI2JU+rsVQNJwYBIwIk3zE0ndyhibmZkVUbakEBF/BFaXOPpxwNyIeDMingMWA/uWKzYzMyuuGucUpkt6NDUv7ZjKBgNLC8apT2VmZlZBlU4KPwU+CIwCVgAXpnIVGTeKzUDSNEkPSnqwM7yH2cysPaloUoiIFyJiU0RsBn7O201E9cCQglFrgOVNzGNWRIyOiNEDBgwob8BmZl1MRZOCpJ0Lev8ZaLgyaQEwSVJPScOA3YC/VDI2MzOD7uWasaRrgbFAf0n1wDnAWEmjyJqGlgCnAkTEE5LmAU8CG4EzImJTuWIzM7PiypYUImJykeIrmhn/XODccsVjZmYt8x3NZmaWc1IwM7Ock4KZmeVKSgqS+kj6ccP9AZIulNSn3MGZmVlllVpTuBJ4FZiQ/l4FflGuoMzMrDpKvfrogxHxqYL+70paWI6AzMysekqtKbwh6aCGHkkHAm+UJyQzM6uWUmsKpwFz0nkEkT399KRyBWVmZtVRUlKIiIXASEnbp/5XyxqVmZlVRalXHw2XNB3YBviRpF9J+nB5QzMzs0or9ZzCNcDuwP1kD6qbB1xerqDMzKw6Sk0KW0XEl4ANEXFFRMxrxbRmZtZBlHqiubekTwLdJf0zWULYvnxhmZlZNZSaFO4Cjk3/x6eyP5YlIjMzq5pSk8IlEfFQWSMxM7OqK/W8gE8qm5l1AaXWFLpL2pHsxrVcRKxu+5DMzKxaSk0KuwN1vDMpBPCBNo/IzMyqptSk8GRE+GY1M7NOzvcamJlZrtSkcEBZozAzs3ah1KRwk6QdGnok7Sjpt2WKyczMqqTUpDAgIl5p6ImIl4GB5QnJzMyqpdSksEnS0IYeSbuQXX1kZmadSKlXH/0bcLeku1L/IcC08oRkZmbVUupLdm6T9BFgf7J7Fb4cESvLGpmZmVVcqS/ZEXAk8JGIuAl4n6R9yxqZmZlVXKnnFH5Cdlnq5NS/FrisLBGZmVnVlHpOYb+I+IikhyG7+kjS1mWMy8zMqqDUmsJbkrqRrjiSNADYXLaozMysKkpNChcDvwYGSjoXuBv4ftmiMjOzqij16qOrJdUBh5JdfXR8RDxV1sjMzKziSkoKkvoCLwLXFpb5fQpmZp1LqSea68jOJwjYGViB36dgZtbplNp8NKyhW9LDfreCmVnn1Kr3KaTLUH0pqplZJ1XqOYWbUucewDXlC8fMzKqp1HMKF5Ddl1AfEc+VMR4zM6uiUpPCYw0d6UokAHz1kZlZ51LqOYWVwBPAg2RXItWl7iZJulLSi5IeLyjrK+l2Sc+k/zumckm6WNJiSY+mJ7KamVmFlZoUpgH1wIXAbhExLCJauhx1NtmTVQudBdwREbsBd6R+gKOA3dLfNOCnJcZlZmZtqKSkEBGXAwcBPYF7JH2mhGn+CDRuXjoOmJO65wDHF5T/MjL3ATtI2rmU2MzMrO2U+j6FTwJHA0vIjuK/KemRd7G8nSJiBUD63/Ce58HA0oLx6lOZmZlVUKknmo9t1F/XxnGoSFnRd0BLmkZ6FejQoUOLjWJmZu9SqXc0f76NlveCpJ0jYkVqHnoxldcDQwrGqwGWNxHLLGAWwOjRo4smDjMze3dKvXltQbHyiBjfyuUtAKYC56X/8wvKp0uaC+wHrGloZjIzs8optfloD+CU1sxY0rXAWKC/pHrgHLJkME/SF4DngRPT6LcC44DFwOtAW9VMzMysFUpNCmsj4q7WzDgiJjcx6NAi4wZwRmvmb2Zmba/U+xRGSnpF0j8kPSTpEkn9yxqZmZlVXKn3KXQD+gIfBCYC/+Dt+w3MzKyTKPnR2RGxOSJei4hnIuJc4LYyxmVmZlVQ6jkFJI0HDkm9d0XEJeUJyczMqqXUO5p/AMwAnkx/Z6YyMzPrREqtKRwNjIqIzQCS5gAPA98qV2BmZlZ5rXkd5w4F3X3aOhAzM6u+UmsKPwAelvQHsucUHQJ8u2xRmZlZVZT67KNrJd0JjCFLCt+MiH+UMzAzM6u8ZpuPJB3d0B0RKyJiQUTMB16T5KuPzMw6mZbOKVyUnlOUk/Rp4FHefsKpmZl1Ei01Hx0M3CJpMDAX+AmwATgsIp4td3Bmban2rFsqurwl5x3d8khm7UyzNYX0+OqPkSWHR4HLI2KcE4KZWefU4iWpEbEWOAqYB3xaUq+yR2VmZlXRbPORpLW8/VpMAdsCqyVtInvi9fZljs/MzCqo2aQQEdtVKhAzM6u+1tzRbGZmnZyTgpmZ5ZwUzMws56RgZmY5JwUzM8s5KZiZWc5JwczMck4KZmaWc1IwM7Ock4KZmeWcFMzMLOekYGZmOScFMzPLOSmYmVnOScHMzHJOCmZmlnNSMDOzXLNvXrN3qj3rlooub8l5R1d0eWZmrimYmVnOScHMzHJOCmZmlnNSMDOznJOCmZnlqnL1kaQlwFpgE7AxIkZL6gtcB9QCS4AJEfFyNeIzM+uqqllT+HhEjIqI0an/LOCOiNgNuCP1m5lZBbWn5qPjgDmpew5wfBVjMTPrkqqVFAL4naQ6SdNS2U4RsQIg/R9YpdjMzLqsat3RfGBELJc0ELhd0tOlTpiSyDSAoUOHlis+M7MuqSo1hYhYnv6/CPwa2Bd4QdLOAOn/i01MOysiRkfE6AEDBlQqZDOzLqHiSUHStpK2a+gGDgceBxYAU9NoU4H5lY7NzKyrq0bz0U7AryU1LP+aiLhN0gPAPElfAJ4HTqxCbGZmXVrFk0JE/A0YWaR8FXBopeMxM7O3tadLUs3MrMqcFMzMLOekYGZmOScFMzPLOSmYmVnO72huz2b2qfDy1lR2eWbW7rimYGZmOdcUzMwa68K1dNcUzMws56RgZmY5JwUzM8s5KZiZWc5JwczMck4KZmaWc1IwM7Ock4KZmeWcFMzMLOekYGZmOScFMzPLOSmYmVnOScHMzHJOCmZmlnNSMDOznJOCmZnlnBTMzCznpGBmZjknBTMzyzkpmJlZzknBzMxy3asdgFmnNbNPhZe3prLLs07JNQUzM8s5KZiZWc5JwczMck4KZmaWc1IwM7Ock4KZmeWcFMzMLOekYGZmOScFMzPLOSmYmVmu3SUFSUdKWiRpsaSzqh2PmVlX0q6SgqRuwGXAUcBwYLKk4dWNysys62hXSQHYF1gcEX+LiA3AXOC4KsdkZtZltLenpA4Glhb01wP7FY4gaRowLfWuk7SoQrFVnKA/sLJiC/yuKraorsD7r+PqAvtul6YGtLekUGzLxDt6ImYBsyoTTnVJejAiRlc7Dnt3vP86rq6879pb81E9MKSgvwZYXqVYzMy6nPaWFB4AdpM0TNLWwCRgQZVjMjPrMtpV81FEbJQ0Hfgt0A24MiKeqHJY1dQlmsk6Me+/jqvL7jtFRMtjmZlZl9Demo/MzKyKnBTMzCznpGBmZjknBbM2IOn/SDpUUu9G5UdWKyYrnaR9JY1J3cMlfUXSuGrHVQ0+0dwBSPp8RPyi2nFYcZLOBM4AngJGATMiYn4a9lBEfKSa8VnzJJ1D9ry17sDtZE9RuBM4DPhtRJxbvegqz0mhA5D0fEQMrXYcVpykx4ADImKdpFrgV8D/RMRFkh6OiA9XNUBrVtp/o4CewD+Amoh4VdI2wP0RsXdVA6ywdnWfQlcm6dGmBgE7VTIWa7VuEbEOICKWSBoL/ErSLhR/dIu1LxsjYhPwuqRnI+JVgIh4Q9LmKsdWcU4K7cdOwBHAy43KBdxT+XCsFf4haVRELARINYZjgCuBvaobmpVgg6T3RcTrwD4NhZL6AE4KVjU3A70bflgKSbqz8uFYK0wBNhYWRMRGYIqkn1UnJGuFQyLiTYCIKEwCPYCp1QmpenxOwczMcr4k1czMck4KZmaWc1KwDkHSuoLunSU9K+nYasZk1hk5KViHImk74FbghxFxU7XjMetsnBSsw5DUA7gRWJBey1o4bJOkhZIWS7o5lR0r6X5JD0v6vaSdUnlvSb+Q9JikRyV9KpUfKekhSY9IuqPI8k+S9FJazsLUfVIa9u+SHpD0uKRZkpTKd03LfiTN+4Op/Btp+Y9IOi+V/UuaxyOSbpD0vlT+vYJxZkr6Wuq+VNJXU/dsSSek7lMkhaT+kmolPS1pTlrXXxXMt2jM1rU5KVhHciXwMeDawkJJ3YDXImIUcErBoLuB/dMdxXOBb6Tys4E1EbFXulv1fyUNAH4OfCoiRgInNhHDdRExKi3ruoLySyNiTETsCWwDHJPKrwYuS/P8KLBC0lHA8cB+qfz8NO6NaR4jyR6Z8YWCeGslnVywzjOAnhFxYaNt0Qv4IvBiQfHuwKy0rq8Cp7cQs3VhTgrWUWwL9AVOAi5rNGwbYH2RaWqA36bHGHwdGJHKDyucR0S8DOwP/DEinktlq1sZ38dTreQx4BPAiNTUNTgifp3muT7dIHUY8IvUXbisPSX9Kc3jMw3xRnbd+Ezgp8DBZDc5/gj4jyJxnAHMAd4oKFsaEX9O3VcBBzUVcyvX2TohJwXrKN4EJkTENcBbkj5TMGwQsLzINJeQHQ3vBZwK9ErlAhrfoFOsrCTp6PwnwAlpWT9Py2qqOaapZc0Gpqd5fLcgXoAfAicDHwSGkNUGzms0/fbAZKDxDXONlxXNxGxdnJOCdRQbI+K11D0dODc9hgBgAvDnItP0AZal7sI7U3+X5gGApB2Be4GPSRqWyvq2IraGH9OV6dHZJwCkZ+jUSzo+zbNnas//HXByQdt+w7K2I2te6kFWU2iI71hgfURcTZY4Lo+IK4EdJH28II4vAxdHxIZG8Q2VdEDqnkzWrFY0ZjMnBetwImIx8Avg++mx1QeSHVk3NhO4XtKfgJUF5d8DdkwnWB8BPh4RLwHTgBtT2XVbzK3peF4hO9J+DPgN8EDB4M8BZ6YHHt4DvD8ibgMWAA9KWgh8LY17NnA/2eObnwZIT+o8l+wHv7EvARdK2jr1i6x5qLGngKkphr7AT1uI2bowP+bCrBNT9ijvm9PJZLMWuaZgZmY51xTMzCznms/xnH0AAAAlSURBVIKZmeWcFMzMLOekYGZmOScFMzPLOSmYmVnOScHMzHL/HzR/9wEMatIhAAAAAElFTkSuQmCC\n",
1668 |       "text/plain": [
1669 |        "<Figure size 432x288 with 1 Axes>"
1670 |       ]
1671 |      },
1672 |      "metadata": {
1673 |       "needs_background": "light"
1674 |      },
1675 |      "output_type": "display_data"
1676 |     }
1677 |    ],
1678 |    "source": [
1679 |     "pd.crosstab(df.Pclass[df.Sex == 'male'], df.Survived).plot(kind='bar')\n",
1680 |     "plt.title('Выживаемость мужчин в разных классах')\n",
1681 |     "plt.xlabel('Класс пассажира')\n",
1682 |     "plt.ylabel('Количество')"
1683 |    ]
1684 |   },
1685 |   {
1686 |    "cell_type": "code",
1687 |    "execution_count": 43,
1688 |    "metadata": {},
1689 |    "outputs": [
1690 |     {
1691 |      "data": {
1692 |       "image/png": "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\n",
1693 |       "text/plain": [
1694 |        "<Figure size 432x288 with 3 Axes>"
1695 |       ]
1696 |      },
1697 |      "metadata": {
1698 |       "needs_background": "light"
1699 |      },
1700 |      "output_type": "display_data"
1701 |     }
1702 |    ],
1703 |    "source": [
1704 |     "# График сопряженности между классом пассажира, полом и переменной выжил/не выжил.\n",
1705 |     "\n",
1706 |     "from statsmodels.graphics.mosaicplot import mosaic\n",
1707 |     "\n",
1708 |     "mosaic(df, ['Pclass', 'Sex', 'Survived'], title='Мозаичный график', horizontal=0, gap=0.03)\n",
1709 |     "plt.show()"
1710 |    ]
1711 |   },
1712 |   {
1713 |    "cell_type": "code",
1714 |    "execution_count": 44,
1715 |    "metadata": {},
1716 |    "outputs": [
1717 |     {
1718 |      "data": {
1719 |       "text/plain": [
1720 |        "27.333333333333332"
1721 |       ]
1722 |      },
1723 |      "execution_count": 44,
1724 |      "metadata": {},
1725 |      "output_type": "execute_result"
1726 |     }
1727 |    ],
1728 |    "source": [
1729 |     "odds_female_pc1 = cross_table_female[1][1] / cross_table_female[1][0]\n",
1730 |     "odds_female_pc1"
1731 |    ]
1732 |   },
1733 |   {
1734 |    "cell_type": "code",
1735 |    "execution_count": 45,
1736 |    "metadata": {},
1737 |    "outputs": [
1738 |     {
1739 |      "data": {
1740 |       "text/plain": [
1741 |        "3.3081069585961433"
1742 |       ]
1743 |      },
1744 |      "execution_count": 45,
1745 |      "metadata": {},
1746 |      "output_type": "execute_result"
1747 |     }
1748 |    ],
1749 |    "source": [
1750 |     "np.log(odds_female_pc1)\n",
1751 |     "# Это значение Intercept"
1752 |    ]
1753 |   },
1754 |   {
1755 |    "cell_type": "code",
1756 |    "execution_count": 46,
1757 |    "metadata": {},
1758 |    "outputs": [
1759 |     {
1760 |      "data": {
1761 |       "text/plain": [
1762 |        "0.6557377049180327"
1763 |       ]
1764 |      },
1765 |      "execution_count": 46,
1766 |      "metadata": {},
1767 |      "output_type": "execute_result"
1768 |     }
1769 |    ],
1770 |    "source": [
1771 |     "odds_male_pc1 = cross_table_male[1][1] / cross_table_male[1][0]\n",
1772 |     "odds_male_pc1"
1773 |    ]
1774 |   },
1775 |   {
1776 |    "cell_type": "code",
1777 |    "execution_count": 47,
1778 |    "metadata": {},
1779 |    "outputs": [
1780 |     {
1781 |      "data": {
1782 |       "text/plain": [
1783 |        "-3.7301013686555184"
1784 |       ]
1785 |      },
1786 |      "execution_count": 47,
1787 |      "metadata": {},
1788 |      "output_type": "execute_result"
1789 |     }
1790 |    ],
1791 |    "source": [
1792 |     "np.log(odds_male_pc1 / odds_female_pc1)\n",
1793 |     "# Это значение C(Sex)[T.male]"
1794 |    ]
1795 |   },
1796 |   {
1797 |    "cell_type": "code",
1798 |    "execution_count": 48,
1799 |    "metadata": {},
1800 |    "outputs": [
1801 |     {
1802 |      "data": {
1803 |       "text/plain": [
1804 |        "11.333333333333334"
1805 |       ]
1806 |      },
1807 |      "execution_count": 48,
1808 |      "metadata": {},
1809 |      "output_type": "execute_result"
1810 |     }
1811 |    ],
1812 |    "source": [
1813 |     "odds_female_pc2 = cross_table_female[2][1] / cross_table_female[2][0]\n",
1814 |     "odds_female_pc2"
1815 |    ]
1816 |   },
1817 |   {
1818 |    "cell_type": "code",
1819 |    "execution_count": 49,
1820 |    "metadata": {},
1821 |    "outputs": [
1822 |     {
1823 |      "data": {
1824 |       "text/plain": [
1825 |        "-0.8803587226480917"
1826 |       ]
1827 |      },
1828 |      "execution_count": 49,
1829 |      "metadata": {},
1830 |      "output_type": "execute_result"
1831 |     }
1832 |    ],
1833 |    "source": [
1834 |     "np.log(odds_female_pc2 / odds_female_pc1)\n",
1835 |     "# Это значение C(Pclass)[T.2]"
1836 |    ]
1837 |   },
1838 |   {
1839 |    "cell_type": "code",
1840 |    "execution_count": 50,
1841 |    "metadata": {},
1842 |    "outputs": [
1843 |     {
1844 |      "data": {
1845 |       "text/plain": [
1846 |        "0.8545454545454545"
1847 |       ]
1848 |      },
1849 |      "execution_count": 50,
1850 |      "metadata": {},
1851 |      "output_type": "execute_result"
1852 |     }
1853 |    ],
1854 |    "source": [
1855 |     "odds_female_pc3 = cross_table_female[3][1] / cross_table_female[3][0]\n",
1856 |     "odds_female_pc3"
1857 |    ]
1858 |   },
1859 |   {
1860 |    "cell_type": "code",
1861 |    "execution_count": 51,
1862 |    "metadata": {},
1863 |    "outputs": [
1864 |     {
1865 |      "data": {
1866 |       "text/plain": [
1867 |        "-3.4652925421185556"
1868 |       ]
1869 |      },
1870 |      "execution_count": 51,
1871 |      "metadata": {},
1872 |      "output_type": "execute_result"
1873 |     }
1874 |    ],
1875 |    "source": [
1876 |     "np.log(odds_female_pc3 / odds_female_pc1)\n",
1877 |     "# Это значение C(Pclass)[T.3]"
1878 |    ]
1879 |   },
1880 |   {
1881 |    "cell_type": "code",
1882 |    "execution_count": 52,
1883 |    "metadata": {},
1884 |    "outputs": [
1885 |     {
1886 |      "data": {
1887 |       "text/plain": [
1888 |        "0.17857142857142858"
1889 |       ]
1890 |      },
1891 |      "execution_count": 52,
1892 |      "metadata": {},
1893 |      "output_type": "execute_result"
1894 |     }
1895 |    ],
1896 |    "source": [
1897 |     "odds_male_pc2 = cross_table_male[2][1] / cross_table_male[2][0]\n",
1898 |     "odds_male_pc2"
1899 |    ]
1900 |   },
1901 |   {
1902 |    "cell_type": "code",
1903 |    "execution_count": 53,
1904 |    "metadata": {},
1905 |    "outputs": [
1906 |     {
1907 |      "data": {
1908 |       "text/plain": [
1909 |        "-0.42041346503363686"
1910 |       ]
1911 |      },
1912 |      "execution_count": 53,
1913 |      "metadata": {},
1914 |      "output_type": "execute_result"
1915 |     }
1916 |    ],
1917 |    "source": [
1918 |     "np.log(odds_male_pc2 / odds_female_pc2) - np.log(odds_male_pc1 / odds_female_pc1)\n",
1919 |     "# Это значение C(Sex)[T.male]:C(Pclass)[T.2]"
1920 |    ]
1921 |   },
1922 |   {
1923 |    "cell_type": "code",
1924 |    "execution_count": 54,
1925 |    "metadata": {},
1926 |    "outputs": [
1927 |     {
1928 |      "data": {
1929 |       "text/plain": [
1930 |        "0.17674418604651163"
1931 |       ]
1932 |      },
1933 |      "execution_count": 54,
1934 |      "metadata": {},
1935 |      "output_type": "execute_result"
1936 |     }
1937 |    ],
1938 |    "source": [
1939 |     "odds_male_pc3 = cross_table_male[3][1] / cross_table_male[3][0]\n",
1940 |     "odds_male_pc3"
1941 |    ]
1942 |   },
1943 |   {
1944 |    "cell_type": "code",
1945 |    "execution_count": 55,
1946 |    "metadata": {},
1947 |    "outputs": [
1948 |     {
1949 |      "data": {
1950 |       "text/plain": [
1951 |        "2.154235083776654"
1952 |       ]
1953 |      },
1954 |      "execution_count": 55,
1955 |      "metadata": {},
1956 |      "output_type": "execute_result"
1957 |     }
1958 |    ],
1959 |    "source": [
1960 |     "np.log(odds_male_pc3 / odds_female_pc3) - np.log(odds_male_pc1 / odds_female_pc1)\n",
1961 |     "# Это значение C(Sex)[T.male]:C(Pclass)[T.3]"
1962 |    ]
1963 |   },
1964 |   {
1965 |    "cell_type": "code",
1966 |    "execution_count": 56,
1967 |    "metadata": {},
1968 |    "outputs": [
1969 |     {
1970 |      "data": {
1971 |       "text/plain": [
1972 |        "-0.42200000000000015"
1973 |       ]
1974 |      },
1975 |      "execution_count": 56,
1976 |      "metadata": {},
1977 |      "output_type": "execute_result"
1978 |     }
1979 |    ],
1980 |    "source": [
1981 |     "# Получаем модель:\n",
1982 |     "# log(odds) = 3.3081 - (3.7301 * Sex_male) - (0.8804 * Pclass2) - (3.4653 * Pclass3) - (0.4204 * Sex_male * Pclass2) + (2.1542 * Sex_male * Pclass3)\n",
1983 |     "\n",
1984 |     "# Если предсказываем логарифм шансов для Ж в 1кл (Sex_male = 0, Pclass2 = 0, Pclass3 = 0):\n",
1985 |     "# log(odds) = 3.3081 - (3.7301 * 0) - (0.8804 * 0) - (3.4653 * 0) - (0.4204 * 0 * 0) + (2.1542 * 0 * 0)\n",
1986 |     "# log(odds) = 3.3081\n",
1987 |     "\n",
1988 |     "# Если предсказываем логарифм шансов для М в 1кл (Sex_male = 1, Pclass2 = 0, Pclass3 = 0):\n",
1989 |     "# log(odds) = 3.3081 - (3.7301 * 1) - (0.8804 * 0) - (3.4653 * 0) - (0.4204 * 1 * 0) + (2.1542 * 1 * 0)\n",
1990 |     "# log(odds) = 3.3081 - 3.7301\n",
1991 |     "# log(odds) = -0.422\n",
1992 |     "\n",
1993 |     "log_odds_male = 3.3081 - 3.7301\n",
1994 |     "log_odds_male"
1995 |    ]
1996 |   },
1997 |   {
1998 |    "cell_type": "code",
1999 |    "execution_count": 57,
2000 |    "metadata": {},
2001 |    "outputs": [
2002 |     {
2003 |      "data": {
2004 |       "text/plain": [
2005 |        "2.4277"
2006 |       ]
2007 |      },
2008 |      "execution_count": 57,
2009 |      "metadata": {},
2010 |      "output_type": "execute_result"
2011 |     }
2012 |    ],
2013 |    "source": [
2014 |     "# Если предсказываем логарифм шансов для Ж во 2кл (Sex_male = 0, Pclass2 = 1, Pclass3 = 0):\n",
2015 |     "# log(odds) = 3.3081 - (3.7301 * 0) - (0.8804 * 1) - (3.4653 * 0) - (0.4204 * 0 * 1) + (2.1542 * 0 * 0)\n",
2016 |     "# log(odds) = 3.3081 - 0.8804\n",
2017 |     "# log(odds) = 2.4277\n",
2018 |     "\n",
2019 |     "log_odds_female_pc2 = 3.3081 - 0.8804\n",
2020 |     "log_odds_female_pc2"
2021 |    ]
2022 |   },
2023 |   {
2024 |    "cell_type": "code",
2025 |    "execution_count": 58,
2026 |    "metadata": {},
2027 |    "outputs": [
2028 |     {
2029 |      "data": {
2030 |       "text/plain": [
2031 |        "-0.1572"
2032 |       ]
2033 |      },
2034 |      "execution_count": 58,
2035 |      "metadata": {},
2036 |      "output_type": "execute_result"
2037 |     }
2038 |    ],
2039 |    "source": [
2040 |     "# Если предсказываем логарифм шансов для Ж в 3кл (Sex_male = 0, Pclass2 = 0, Pclass3 = 1):\n",
2041 |     "# log(odds) = 3.3081 - (3.7301 * 0) - (0.8804 * 0) - (3.4653 * 1) - (0.4204 * 0 * 0) + (2.1542 * 0 * 1)\n",
2042 |     "# log(odds) = 3.3081 - 3.4653\n",
2043 |     "# log(odds) = -0.1572\n",
2044 |     "\n",
2045 |     "log_odds_female_pc3 = 3.3081 - 3.4653\n",
2046 |     "log_odds_female_pc3"
2047 |    ]
2048 |   },
2049 |   {
2050 |    "cell_type": "code",
2051 |    "execution_count": 59,
2052 |    "metadata": {},
2053 |    "outputs": [
2054 |     {
2055 |      "data": {
2056 |       "text/plain": [
2057 |        "-1.7227999999999999"
2058 |       ]
2059 |      },
2060 |      "execution_count": 59,
2061 |      "metadata": {},
2062 |      "output_type": "execute_result"
2063 |     }
2064 |    ],
2065 |    "source": [
2066 |     "# Если предсказываем логарифм шансов для М во 2кл (Sex_male = 1, Pclass2 = 1, Pclass3 = 0):\n",
2067 |     "# log(odds) = 3.3081 - (3.7301 * 1) - (0.8804 * 1) - (3.4653 * 0) - (0.4204 * 1 * 1) + (2.1542 * 1 * 0)\n",
2068 |     "# log(odds) = 3.3081 - 3.7301 - 0.8804 - 0.4204\n",
2069 |     "# log(odds) = -1.7228\n",
2070 |     "\n",
2071 |     "log_odds_male_pc2 = 3.3081 - 3.7301 - 0.8804 - 0.4204\n",
2072 |     "log_odds_male_pc2"
2073 |    ]
2074 |   },
2075 |   {
2076 |    "cell_type": "code",
2077 |    "execution_count": 60,
2078 |    "metadata": {},
2079 |    "outputs": [
2080 |     {
2081 |      "data": {
2082 |       "text/plain": [
2083 |        "-1.7331000000000003"
2084 |       ]
2085 |      },
2086 |      "execution_count": 60,
2087 |      "metadata": {},
2088 |      "output_type": "execute_result"
2089 |     }
2090 |    ],
2091 |    "source": [
2092 |     "# Если предсказываем логарифм шансов для М в 3кл (Sex_male = 1, Pclass2 = 0, Pclass3 = 1):\n",
2093 |     "# log(odds) = 3.3081 - (3.7301 * 1) - (0.8804 * 0) - (3.4653 * 1) - (0.4204 * 1 * 0) + (2.1542 * 1 * 1)\n",
2094 |     "# log(odds) = 3.3081 - 3.7301 - 3.4653 + 2.1542\n",
2095 |     "# log(odds) = -1.7331\n",
2096 |     "\n",
2097 |     "log_odds_male_pc3 = 3.3081 - 3.7301 - 3.4653 + 2.1542\n",
2098 |     "log_odds_male_pc3"
2099 |    ]
2100 |   },
2101 |   {
2102 |    "cell_type": "code",
2103 |    "execution_count": 61,
2104 |    "metadata": {},
2105 |    "outputs": [
2106 |     {
2107 |      "data": {
2108 |       "text/plain": [
2109 |        "654.2750443468178"
2110 |       ]
2111 |      },
2112 |      "execution_count": 61,
2113 |      "metadata": {},
2114 |      "output_type": "execute_result"
2115 |     }
2116 |    ],
2117 |    "source": [
2118 |     "# aic у этой модели ниже, чем у двух предыдущих, значит эта модель лучше.\n",
2119 |     "glm_binomial.aic"
2120 |    ]
2121 |   },
2122 |   {
2123 |    "cell_type": "markdown",
2124 |    "metadata": {},
2125 |    "source": [
2126 |     "## Комбинация предикторов разного типа"
2127 |    ]
2128 |   },
2129 |   {
2130 |    "cell_type": "code",
2131 |    "execution_count": 62,
2132 |    "metadata": {},
2133 |    "outputs": [
2134 |     {
2135 |      "data": {
2136 |       "text/html": [
2137 |        "<table class=\"simpletable\">\n",
2138 |        "<caption>Generalized Linear Model Regression Results</caption>\n",
2139 |        "<tr>\n",
2140 |        "  <th>Dep. Variable:</th>       <td>Survived</td>     <th>  No. Observations:  </th>  <td>   714</td> \n",
2141 |        "</tr>\n",
2142 |        "<tr>\n",
2143 |        "  <th>Model:</th>                  <td>GLM</td>       <th>  Df Residuals:      </th>  <td>   709</td> \n",
2144 |        "</tr>\n",
2145 |        "<tr>\n",
2146 |        "  <th>Model Family:</th>        <td>Binomial</td>     <th>  Df Model:          </th>  <td>     4</td> \n",
2147 |        "</tr>\n",
2148 |        "<tr>\n",
2149 |        "  <th>Link Function:</th>         <td>logit</td>      <th>  Scale:             </th> <td>  1.0000</td>\n",
2150 |        "</tr>\n",
2151 |        "<tr>\n",
2152 |        "  <th>Method:</th>                <td>IRLS</td>       <th>  Log-Likelihood:    </th> <td> -323.64</td>\n",
2153 |        "</tr>\n",
2154 |        "<tr>\n",
2155 |        "  <th>Date:</th>            <td>Tue, 08 Jun 2021</td> <th>  Deviance:          </th> <td>  647.28</td>\n",
2156 |        "</tr>\n",
2157 |        "<tr>\n",
2158 |        "  <th>Time:</th>                <td>20:53:35</td>     <th>  Pearson chi2:      </th>  <td>  767.</td> \n",
2159 |        "</tr>\n",
2160 |        "<tr>\n",
2161 |        "  <th>No. Iterations:</th>          <td>5</td>        <th>                     </th>     <td> </td>   \n",
2162 |        "</tr>\n",
2163 |        "<tr>\n",
2164 |        "  <th>Covariance Type:</th>     <td>nonrobust</td>    <th>                     </th>     <td> </td>   \n",
2165 |        "</tr>\n",
2166 |        "</table>\n",
2167 |        "<table class=\"simpletable\">\n",
2168 |        "<tr>\n",
2169 |        "         <td></td>           <th>coef</th>     <th>std err</th>      <th>z</th>      <th>P>|z|</th>  <th>[0.025</th>    <th>0.975]</th>  \n",
2170 |        "</tr>\n",
2171 |        "<tr>\n",
2172 |        "  <th>Intercept</th>      <td>    3.7770</td> <td>    0.401</td> <td>    9.416</td> <td> 0.000</td> <td>    2.991</td> <td>    4.563</td>\n",
2173 |        "</tr>\n",
2174 |        "<tr>\n",
2175 |        "  <th>C(Sex)[T.male]</th> <td>   -2.5228</td> <td>    0.207</td> <td>  -12.164</td> <td> 0.000</td> <td>   -2.929</td> <td>   -2.116</td>\n",
2176 |        "</tr>\n",
2177 |        "<tr>\n",
2178 |        "  <th>C(Pclass)[T.2]</th> <td>   -1.3098</td> <td>    0.278</td> <td>   -4.710</td> <td> 0.000</td> <td>   -1.855</td> <td>   -0.765</td>\n",
2179 |        "</tr>\n",
2180 |        "<tr>\n",
2181 |        "  <th>C(Pclass)[T.3]</th> <td>   -2.5806</td> <td>    0.281</td> <td>   -9.169</td> <td> 0.000</td> <td>   -3.132</td> <td>   -2.029</td>\n",
2182 |        "</tr>\n",
2183 |        "<tr>\n",
2184 |        "  <th>Age</th>            <td>   -0.0370</td> <td>    0.008</td> <td>   -4.831</td> <td> 0.000</td> <td>   -0.052</td> <td>   -0.022</td>\n",
2185 |        "</tr>\n",
2186 |        "</table>"
2187 |       ],
2188 |       "text/plain": [
2189 |        "<class 'statsmodels.iolib.summary.Summary'>\n",
2190 |        "\"\"\"\n",
2191 |        "                 Generalized Linear Model Regression Results                  \n",
2192 |        "==============================================================================\n",
2193 |        "Dep. Variable:               Survived   No. Observations:                  714\n",
2194 |        "Model:                            GLM   Df Residuals:                      709\n",
2195 |        "Model Family:                Binomial   Df Model:                            4\n",
2196 |        "Link Function:                  logit   Scale:                          1.0000\n",
2197 |        "Method:                          IRLS   Log-Likelihood:                -323.64\n",
2198 |        "Date:                Tue, 08 Jun 2021   Deviance:                       647.28\n",
2199 |        "Time:                        20:53:35   Pearson chi2:                     767.\n",
2200 |        "No. Iterations:                     5                                         \n",
2201 |        "Covariance Type:            nonrobust                                         \n",
2202 |        "==================================================================================\n",
2203 |        "                     coef    std err          z      P>|z|      [0.025      0.975]\n",
2204 |        "----------------------------------------------------------------------------------\n",
2205 |        "Intercept          3.7770      0.401      9.416      0.000       2.991       4.563\n",
2206 |        "C(Sex)[T.male]    -2.5228      0.207    -12.164      0.000      -2.929      -2.116\n",
2207 |        "C(Pclass)[T.2]    -1.3098      0.278     -4.710      0.000      -1.855      -0.765\n",
2208 |        "C(Pclass)[T.3]    -2.5806      0.281     -9.169      0.000      -3.132      -2.029\n",
2209 |        "Age               -0.0370      0.008     -4.831      0.000      -0.052      -0.022\n",
2210 |        "==================================================================================\n",
2211 |        "\"\"\""
2212 |       ]
2213 |      },
2214 |      "execution_count": 62,
2215 |      "metadata": {},
2216 |      "output_type": "execute_result"
2217 |     }
2218 |    ],
2219 |    "source": [
2220 |     "glm_binomial = smf.glm(formula='Survived ~ C(Sex) + C(Pclass) + Age', data=df, family=sm.families.Binomial()).fit()\n",
2221 |     "glm_binomial.summary()\n",
2222 |     "\n",
2223 |     "# Intercept - логарифм шансов для Ж в 1кл при условии, что их возраст равен нулю.\n",
2224 |     "# Age - коэффициент при возрасте показывает,\n",
2225 |     "# на сколько изменяется логарифм шансов для базового уровня при единичном изменении возраста."
2226 |    ]
2227 |   },
2228 |   {
2229 |    "cell_type": "code",
2230 |    "execution_count": 63,
2231 |    "metadata": {},
2232 |    "outputs": [
2233 |     {
2234 |      "data": {
2235 |       "text/plain": [
2236 |        "3.037"
2237 |       ]
2238 |      },
2239 |      "execution_count": 63,
2240 |      "metadata": {},
2241 |      "output_type": "execute_result"
2242 |     }
2243 |    ],
2244 |    "source": [
2245 |     "# Получаем модель:\n",
2246 |     "# log(odds) = 3.7770 - (2.5228 * Sex_male) - (1.3098 * Pclass2) - (2.5806 * Pclass3) - (0.0370 * Age)\n",
2247 |     "\n",
2248 |     "# Если предсказываем логарифм шансов для Ж в 1кл в возрасте 20 лет:\n",
2249 |     "# log(odds) = 3.7770 - (2.5228 * 0) - (1.3098 * 0) - (2.5806 * 0) - (0.0370 * 20)\n",
2250 |     "# log(odds) = 3.7770 - (0.0370 * 20)\n",
2251 |     "# log(odds) = 3.037\n",
2252 |     "\n",
2253 |     "3.7770 - (0.0370 * 20)"
2254 |    ]
2255 |   },
2256 |   {
2257 |    "cell_type": "code",
2258 |    "execution_count": 64,
2259 |    "metadata": {},
2260 |    "outputs": [
2261 |     {
2262 |      "data": {
2263 |       "text/plain": [
2264 |        "657.2831255018241"
2265 |       ]
2266 |      },
2267 |      "execution_count": 64,
2268 |      "metadata": {},
2269 |      "output_type": "execute_result"
2270 |     }
2271 |    ],
2272 |    "source": [
2273 |     "# Кстати, модель без учета возраста была лучше.\n",
2274 |     "glm_binomial.aic"
2275 |    ]
2276 |   },
2277 |   {
2278 |    "cell_type": "code",
2279 |    "execution_count": null,
2280 |    "metadata": {},
2281 |    "outputs": [],
2282 |    "source": []
2283 |   }
2284 |  ],
2285 |  "metadata": {
2286 |   "kernelspec": {
2287 |    "display_name": "Python 3",
2288 |    "language": "python",
2289 |    "name": "python3"
2290 |   },
2291 |   "language_info": {
2292 |    "codemirror_mode": {
2293 |     "name": "ipython",
2294 |     "version": 3
2295 |    },
2296 |    "file_extension": ".py",
2297 |    "mimetype": "text/x-python",
2298 |    "name": "python",
2299 |    "nbconvert_exporter": "python",
2300 |    "pygments_lexer": "ipython3",
2301 |    "version": "3.7.4"
2302 |   }
2303 |  },
2304 |  "nbformat": 4,
2305 |  "nbformat_minor": 2
2306 | }
2307 | 


--------------------------------------------------------------------------------