LDA, QDA.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Читаем данные"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>feature1</th>\n",
       "      <th>feature2</th>\n",
       "      <th>feature3</th>\n",
       "      <th>feature4</th>\n",
       "      <th>feature5</th>\n",
       "      <th>feature6</th>\n",
       "      <th>feature7</th>\n",
       "      <th>feature8</th>\n",
       "      <th>feature9</th>\n",
       "      <th>feature10</th>\n",
       "      <th>...</th>\n",
       "      <th>feature20</th>\n",
       "      <th>feature21</th>\n",
       "      <th>feature22</th>\n",
       "      <th>feature23</th>\n",
       "      <th>feature24</th>\n",
       "      <th>feature25</th>\n",
       "      <th>feature26</th>\n",
       "      <th>feature27</th>\n",
       "      <th>feature28</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 29 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   feature1  feature2  feature3  feature4  feature5  feature6  feature7  \\\n",
       "0         0         1         2         2         3         1         1   \n",
       "1         3         3         2         1         1         1         1   \n",
       "2         2         2         2         1         3         2         1   \n",
       "3         2         3         2         1         2         2         2   \n",
       "4         3         3         3         2         3         3         3   \n",
       "\n",
       "   feature8  feature9  feature10  ...    feature20  feature21  feature22  \\\n",
       "0         1         2          2  ...            0          1          1   \n",
       "1         2         3          3  ...            1          1          1   \n",
       "2         2         1          3  ...            1          1          0   \n",
       "3         2         3          3  ...            1          1          0   \n",
       "4         3         2          2  ...            0          0          1   \n",
       "\n",
       "   feature23  feature24  feature25  feature26  feature27  feature28  label  \n",
       "0          1          1          1          1          0          1      1  \n",
       "1          1          0          0          1          0          0      1  \n",
       "2          1          0          0          1          1          1      1  \n",
       "3          1          1          1          1          1          0      1  \n",
       "4          1          1          1          1          1          1      1  \n",
       "\n",
       "[5 rows x 29 columns]"
      ]
     },
     "execution_count": 93,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "%matplotlib inline\n",
    "data = pd.read_csv('numeric sequence.csv', sep = ',', engine = 'python')\n",
    "processed_data = data\n",
    "for i in range (1, 15):\n",
    "     processed_data['feature'+str(i)] = data['feature'+str(i*2-1)]*2 + data['feature'+str(i*2)]\n",
    "to_drop=[]\n",
    "for i in range (15, 29):\n",
    "    to_drop.append('feature'+str(i))\n",
    "# processed_data = processed_data.drop(to_drop, 1).drop('label', 1)\n",
    "# processed_data = processed_data.replace(0, 'A')\\\n",
    "#                               .replace(1, 'T')\\\n",
    "#                               .replace(2, 'C')\\\n",
    "#                               .replace(3, 'G')\n",
    "# processed_data = pd.get_dummies(processed_data)\n",
    "# processed_data=pd.concat((processed_data, data['label']), axis = 1)\n",
    "data = processed_data\n",
    "data.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Разобьем данные на тренеровочные и тестовые"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "X, X_test, y, y_test = train_test_split(data.drop('label', axis=1), data['label'], test_size = 0.33, \n",
    "                                                    random_state = 2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    " # Линейный дискриминантный анализ"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Programs\\Anaconda3\\lib\\site-packages\\sklearn\\discriminant_analysis.py:388: UserWarning: Variables are collinear.\n",
      "  warnings.warn(\"Variables are collinear.\")\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "LinearDiscriminantAnalysis(n_components=None, priors=None, shrinkage=None,\n",
       "              solver='svd', store_covariance=False, tol=0.0001)"
      ]
     },
     "execution_count": 95,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn import discriminant_analysis\n",
    "lda_model = discriminant_analysis.LinearDiscriminantAnalysis()\n",
    "lda_model.fit(X.values, y.values)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Смотрим, как точно получилось"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.151119402985 0.151515151515\n"
     ]
    }
   ],
   "source": [
    "err_train = np.mean(y != lda_model.predict(X)) # Ошибка на обучающей выборке\n",
    "err_test = np.mean(y_test != lda_model.predict(X_test)) # Ошибка на тестовой выборке\n",
    "print (err_train, err_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "$\n",
    "\\begin{array}{|c|c|}\n",
    "\\hline\n",
    "{\\rm TN} & {\\rm FN} \\\\\n",
    "\\hline\n",
    "{\\rm FP} & {\\rm TP} \\\\\n",
    "\\hline\n",
    "\\end{array}\n",
    "$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[680 121]\n",
      " [122 685]]\n",
      "[[334  56]\n",
      " [ 64 338]]\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "print(confusion_matrix(lda_model.predict(X), y)) # confusion matrix for train sample\n",
    "print(confusion_matrix(lda_model.predict(X_test), y_test)) # confusion matrix for test sample"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## ROC-кривая"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import roc_curve\n",
    "proba = lda_model.predict_proba(X_test)[:, 1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1649    False\n",
       "1387    False\n",
       "1766    False\n",
       "1805    False\n",
       "1861    False\n",
       "1504    False\n",
       "954      True\n",
       "990      True\n",
       "1998    False\n",
       "1066     True\n",
       "236      True\n",
       "2337    False\n",
       "1353    False\n",
       "2264    False\n",
       "981      True\n",
       "79       True\n",
       "1719    False\n",
       "62       True\n",
       "209      True\n",
       "904      True\n",
       "643      True\n",
       "230      True\n",
       "370      True\n",
       "431      True\n",
       "112      True\n",
       "2345    False\n",
       "95       True\n",
       "1061     True\n",
       "1264    False\n",
       "540      True\n",
       "        ...  \n",
       "271      True\n",
       "343      True\n",
       "277      True\n",
       "465      True\n",
       "305      True\n",
       "1068     True\n",
       "1124     True\n",
       "734      True\n",
       "960      True\n",
       "367      True\n",
       "2348    False\n",
       "759      True\n",
       "2180    False\n",
       "757      True\n",
       "2198    False\n",
       "85       True\n",
       "1325    False\n",
       "2298    False\n",
       "116      True\n",
       "1574    False\n",
       "34       True\n",
       "352      True\n",
       "2352    False\n",
       "1324    False\n",
       "2039    False\n",
       "1415    False\n",
       "1851    False\n",
       "7        True\n",
       "2273    False\n",
       "917      True\n",
       "Name: label, Length: 792, dtype: bool"
      ]
     },
     "execution_count": 118,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_test == 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 121,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXoAAAD8CAYAAAB5Pm/hAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nfwCr+b0ay5CbSwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2a7f51f5f60>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.metrics import roc_curve\n",
    "proba = lda_model.predict_proba(X_test)[:, 1]\n",
    "plt.hist(proba[y_test == 1], color = 'b', alpha = 0.5, normed = False)\n",
    "plt.hist(proba[y_test == 0], color = 'r', alpha = 0.5, normed = False)\n",
    "pass"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[-0.05, 1.05, -0.05, 1.05]"
      ]
     },
     "execution_count": 100,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n55xzuTwpOOecy/X/bo1Uu43t5YIAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2a7f4e9acc0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fpr, tpr, thresholds = roc_curve(y_test, proba)\n",
    "plt.title('Receiver Operating Characteristic')\n",
    "plt.plot(np.concatenate(([0], fpr)), np.concatenate(([0], tpr)))\n",
    "#plt.plot(fpr, tpr)\n",
    "plt.xlabel('False Positive Rate (1 - Specificity)')\n",
    "plt.ylabel('True Positive Rate (Sensitivity)')\n",
    "plt.plot([0, 1], [0, 1], 'k--')\n",
    "plt.axis([-0.05, 1.05, -0.05, 1.05])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Перекрестный контроль"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ 0.85185185  0.82098765  0.86956522  0.80745342  0.85714286  0.8757764\n",
      "  0.83125     0.85        0.83125     0.81875   ]\n",
      "Accuracy: 0.84 (+/- 0.04)\n",
      "Error:    0.16 (+/- 0.04)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Programs\\Anaconda3\\lib\\site-packages\\sklearn\\discriminant_analysis.py:388: UserWarning: Variables are collinear.\n",
      "  warnings.warn(\"Variables are collinear.\")\n",
      "C:\\Programs\\Anaconda3\\lib\\site-packages\\sklearn\\discriminant_analysis.py:388: UserWarning: Variables are collinear.\n",
      "  warnings.warn(\"Variables are collinear.\")\n",
      "C:\\Programs\\Anaconda3\\lib\\site-packages\\sklearn\\discriminant_analysis.py:388: UserWarning: Variables are collinear.\n",
      "  warnings.warn(\"Variables are collinear.\")\n",
      "C:\\Programs\\Anaconda3\\lib\\site-packages\\sklearn\\discriminant_analysis.py:388: UserWarning: Variables are collinear.\n",
      "  warnings.warn(\"Variables are collinear.\")\n",
      "C:\\Programs\\Anaconda3\\lib\\site-packages\\sklearn\\discriminant_analysis.py:388: UserWarning: Variables are collinear.\n",
      "  warnings.warn(\"Variables are collinear.\")\n",
      "C:\\Programs\\Anaconda3\\lib\\site-packages\\sklearn\\discriminant_analysis.py:388: UserWarning: Variables are collinear.\n",
      "  warnings.warn(\"Variables are collinear.\")\n",
      "C:\\Programs\\Anaconda3\\lib\\site-packages\\sklearn\\discriminant_analysis.py:388: UserWarning: Variables are collinear.\n",
      "  warnings.warn(\"Variables are collinear.\")\n",
      "C:\\Programs\\Anaconda3\\lib\\site-packages\\sklearn\\discriminant_analysis.py:388: UserWarning: Variables are collinear.\n",
      "  warnings.warn(\"Variables are collinear.\")\n",
      "C:\\Programs\\Anaconda3\\lib\\site-packages\\sklearn\\discriminant_analysis.py:388: UserWarning: Variables are collinear.\n",
      "  warnings.warn(\"Variables are collinear.\")\n",
      "C:\\Programs\\Anaconda3\\lib\\site-packages\\sklearn\\discriminant_analysis.py:388: UserWarning: Variables are collinear.\n",
      "  warnings.warn(\"Variables are collinear.\")\n"
     ]
    }
   ],
   "source": [
    "from sklearn import cross_validation\n",
    "model = discriminant_analysis.LinearDiscriminantAnalysis()\n",
    "scores = cross_validation.cross_val_score(model, X, y, cv = 10)\n",
    "print (scores)\n",
    "print(\"Accuracy: %0.2f (+/- %0.2f)\" % (scores.mean(), scores.std() * 2))\n",
    "print(\"Error:    %0.2f (+/- %0.2f)\" % (1 - scores.mean(), scores.std() * 2))\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Точность не выросла. Для этого набора данных метод перекрестного контроля не так уж и нужен, т. к. кажется, что данных достаточно много."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Квадратичный дискриминантный анализ"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Programs\\Anaconda3\\lib\\site-packages\\sklearn\\discriminant_analysis.py:682: UserWarning: Variables are collinear\n",
      "  warnings.warn(\"Variables are collinear\")\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "QuadraticDiscriminantAnalysis(priors=None, reg_param=0.0,\n",
       "               store_covariance=False, store_covariances=None, tol=0.0001)"
      ]
     },
     "execution_count": 102,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "qda = discriminant_analysis.QuadraticDiscriminantAnalysis()\n",
    "qda.fit(X.values, y.values)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.410447761194 0.424242424242\n"
     ]
    }
   ],
   "source": [
    "err_train = np.mean(y != qda.predict(X)) # Ошибка на обучающей выборке\n",
    "err_test = np.mean(y_test != qda.predict(X_test)) # Ошибка на тестовой выборке\n",
    "print (err_train, err_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "$\n",
    "\\begin{array}{|c|c|}\n",
    "\\hline\n",
    "{\\rm TN} & {\\rm FN} \\\\\n",
    "\\hline\n",
    "{\\rm FP} & {\\rm TP} \\\\\n",
    "\\hline\n",
    "\\end{array}\n",
    "$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[488 346]\n",
      " [314 460]]\n",
      "[[252 190]\n",
      " [146 204]]\n"
     ]
    }
   ],
   "source": [
    "print(confusion_matrix(qda.predict(X), y)) # confusion matrix for train sample\n",
    "print(confusion_matrix(qda.predict(X_test), y_test)) # confusion matrix for test sample"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## ROC-кривая"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXoAAAD8CAYAAAB5Pm/hAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2a7f4e9a828>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.metrics import roc_curve\n",
    "proba = qda.predict_proba(X_test)[:, 1]\n",
    "plt.hist(proba[y_test == 0], color = 'b', alpha = 0.5, normed = False)\n",
    "plt.hist(proba[y_test == 1], color = 'r', alpha = 0.5, normed = False)\n",
    "pass"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[-0.05, 1.05, -0.05, 1.05]"
      ]
     },
     "execution_count": 106,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2a7f5106668>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fpr, tpr, thresholds = roc_curve(y_test, proba)\n",
    "plt.title('Receiver Operating Characteristic')\n",
    "plt.plot(np.concatenate(([0], fpr)), np.concatenate(([0], tpr)))\n",
    "#plt.plot(fpr, tpr)\n",
    "plt.xlabel('False Positive Rate (1 - Specificity)')\n",
    "plt.ylabel('True Positive Rate (Sensitivity)')\n",
    "plt.plot([0, 1], [0, 1], 'k--')\n",
    "plt.axis([-0.05, 1.05, -0.05, 1.05])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "На самом деле эти методы не совсем применимы к набору данных, так как все признаки категориальные. И, применяя рассмотренные методы, мы вводим способ сравнения аденина, гуанина, цитозина и тимина."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "kNN давал результаты лучше"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}