Benchmark_DMIA.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Подготовил Семененко Денис, 24.10.15"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# DMIA 2015 Fall Task 1 - Classification \n",
    "Определение слов, являющихся фамилиями \n",
    "\n",
    "https://inclass.kaggle.com/c/dmia-2015-fall-task-1-classification"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import sys\n",
    "reload(sys)\n",
    "sys.setdefaultencoding(\"utf-8\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "from pandas import DataFrame as df\n",
    "from pandas import read_csv\n",
    "\n",
    "import sklearn\n",
    "\n",
    "import matplotlib.pyplot as pl\n",
    "%matplotlib inline\n",
    "import seaborn as sns\n",
    "\n",
    "import itertools\n",
    "from sklearn.metrics import roc_auc_score"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##Загрузка данных"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "!mkdir data\n",
    "\n",
    "!cd data && wget 'https://kaggle2.blob.core.windows.net/competitions-data/inclass/4677/linear_test.txt?sv=2012-02-12&se=2015-10-26T21%3A07%3A32Z&sr=b&sp=r&sig=ccIpZtIvNPgDJ60B5mbtHBEgcw9euRjewzi3UWvJP0Q%3D' -O test.csv\n",
    "!cd data && wget 'https://kaggle2.blob.core.windows.net/competitions-data/inclass/4677/linear_train.txt?sv=2012-02-12&se=2015-10-26T21%3A08%3A48Z&sr=b&sp=r&sig=OLAJoRzsAq2%2BMSFpLSMDYX9WCxx%2F1ej%2F7sCGlBYrfZM%3D' -O train.csv"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Word</th>\n",
       "      <th>Label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Аалтонен</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Аар</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Аарон</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>ААРОН</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Аарона</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Word  Label\n",
       "0  Аалтонен      1\n",
       "1       Аар      0\n",
       "2     Аарон      0\n",
       "3     ААРОН      0\n",
       "4    Аарона      0"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train = read_csv('data/train.csv', names=['Word', 'Label'])\n",
    "train[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "101408"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##Пара методов для работы со строками"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "16"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "str = 'Аалтонен'\n",
    "len(str)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Так как наши данные на русском языке, в кодировке они занимают 2 байта. Для того, чтобы адекватно работать с символами можно применять decode('utf-8')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "8"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(str.decode('utf-8'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Привести к нижнему регистру"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'dmia'"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "'DMiA'.lower()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[u'\\u0410',\n",
       " u'\\u0430',\n",
       " u'\\u043b',\n",
       " u'\\u0442',\n",
       " u'\\u043e',\n",
       " u'\\u043d',\n",
       " u'\\u0435',\n",
       " u'\\u043d']"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#сделать лист букв\n",
    "[letter for letter in str.decode('utf-8')]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('D', 'M'), ('M', 'I'), ('I', 'A')]"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#пройти по слову окном в две буквы\n",
    "l = [letter for letter in 'DMIA']\n",
    "[(x,y) for x,y in itertools.izip(l, l[1:])]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "u'\\u043d\\u0435\\u043d'"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#взять 3 последних символа в строке\n",
    "str.decode('utf-8')[-3:]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Все можно найти здесь https://docs.python.org/2/library/string.html или в гугле"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##Feature Engineering"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train['Word'] = train['Word'].apply(lambda word: word.decode('utf-8'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Начнем с общих характеристик \n",
    "\n",
    "Добавим длину слова"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Word</th>\n",
       "      <th>Label</th>\n",
       "      <th>Length</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Аалтонен</td>\n",
       "      <td>1</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Аар</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Аарон</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>ААРОН</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Аарона</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Word  Label  Length\n",
       "0  Аалтонен      1       8\n",
       "1       Аар      0       3\n",
       "2     Аарон      0       5\n",
       "3     ААРОН      0       5\n",
       "4    Аарона      0       6"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train['Length'] = train['Word'].apply(lambda word: len(word))\n",
    "train[:5]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Добавим количество гласных и согласных"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Word</th>\n",
       "      <th>Label</th>\n",
       "      <th>Length</th>\n",
       "      <th>Vowels</th>\n",
       "      <th>Consonants</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Аалтонен</td>\n",
       "      <td>1</td>\n",
       "      <td>8</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Аар</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Аарон</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>ААРОН</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Аарона</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Word  Label  Length  Vowels  Consonants\n",
       "0  Аалтонен      1       8       4           4\n",
       "1       Аар      0       3       2           1\n",
       "2     Аарон      0       5       3           2\n",
       "3     ААРОН      0       5       3           2\n",
       "4    Аарона      0       6       4           2"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "vowels = [ 'а',  'я', 'ё', 'у','е', 'о', 'э', 'ю', 'и', 'ы', 'Ё', 'У', 'Е', 'Ы','А', 'О', 'Э', 'Ю', 'И', 'Я' ]\n",
    "vowels = [let.decode('utf-8') for let in vowels]\n",
    "train['Vowels'] = train['Word'].apply(lambda word: len([letter for letter in word if letter in vowels]))\n",
    "train['Consonants'] = train['Length'] - train['Vowels']\n",
    "train[:5]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Теперь попробуем придумать какие-нибудь критерии и сделаем на их основе фичи"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Например, гипотеза: слова, начинающиеся с маленькой буквы не являются фамилиями.\n",
    "\n",
    "Добавим такой признак"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Word</th>\n",
       "      <th>Label</th>\n",
       "      <th>Length</th>\n",
       "      <th>Vowels</th>\n",
       "      <th>Consonants</th>\n",
       "      <th>is_lower</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>Аахена</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>Абабков</td>\n",
       "      <td>1</td>\n",
       "      <td>7</td>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>абажур</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>абажуром</td>\n",
       "      <td>0</td>\n",
       "      <td>8</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>абажуры</td>\n",
       "      <td>0</td>\n",
       "      <td>7</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        Word  Label  Length  Vowels  Consonants  is_lower\n",
       "8     Аахена      0       6       4           2         0\n",
       "9    Абабков      1       7       3           4         0\n",
       "10    абажур      0       6       3           3         1\n",
       "11  абажуром      0       8       4           4         1\n",
       "12   абажуры      0       7       4           3         1"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train['is_lower'] = train['Word'].apply(lambda word: int( word[0] >= 'а'.decode('utf-8')))\n",
    "train[8:13]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<seaborn.axisgrid.JointGrid at 0x7f01c3a5ce10>"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAa0AAAGoCAYAAAD1m7qEAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAADGoLQAAMagtAAAxvhfqFIDxty9660AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f01c4535210>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.jointplot(x = 'is_lower', y = 'Label', data = train, size = 6)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Вторая гипотеза: сдвоенные буквы встречаются преимущественно не в фамилиях"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Word</th>\n",
       "      <th>Label</th>\n",
       "      <th>Length</th>\n",
       "      <th>Vowels</th>\n",
       "      <th>Consonants</th>\n",
       "      <th>is_lower</th>\n",
       "      <th>Double</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>Ааронов</td>\n",
       "      <td>0</td>\n",
       "      <td>7</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>Аахена</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>Абабков</td>\n",
       "      <td>1</td>\n",
       "      <td>7</td>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>абажур</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>абажуром</td>\n",
       "      <td>0</td>\n",
       "      <td>8</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        Word  Label  Length  Vowels  Consonants  is_lower  Double\n",
       "7    Ааронов      0       7       4           3         0       1\n",
       "8     Аахена      0       6       4           2         0       1\n",
       "9    Абабков      1       7       3           4         0       0\n",
       "10    абажур      0       6       3           3         1       0\n",
       "11  абажуром      0       8       4           4         1       0"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def countDoubles(word):\n",
    "    l = [let for let in word.lower()]\n",
    "    return len([(x,y) for x,y in itertools.izip(l, l[1:]) if x == y])\n",
    "    \n",
    "train['Double'] = train['Word'].apply(lambda word: countDoubles(word))\n",
    "train[7:12]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<seaborn.axisgrid.JointGrid at 0x7f01c3916590>"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAa0AAAGoCAYAAAD1m7qEAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\npQUAMAalBQAwBqUFADDG/wGsGBR9LBOqdAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f01c3965590>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.jointplot(x = 'Double', y = 'Label', data = train, size = 6)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Интересные слова"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Word</th>\n",
       "      <th>Label</th>\n",
       "      <th>Length</th>\n",
       "      <th>Vowels</th>\n",
       "      <th>Consonants</th>\n",
       "      <th>is_lower</th>\n",
       "      <th>Double</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>11900</th>\n",
       "      <td>Вимм-Билль-Данн Продукты Питания</td>\n",
       "      <td>0</td>\n",
       "      <td>32</td>\n",
       "      <td>10</td>\n",
       "      <td>22</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12699</th>\n",
       "      <td>военнообязанный</td>\n",
       "      <td>0</td>\n",
       "      <td>15</td>\n",
       "      <td>7</td>\n",
       "      <td>8</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12700</th>\n",
       "      <td>военнообязанными</td>\n",
       "      <td>0</td>\n",
       "      <td>16</td>\n",
       "      <td>8</td>\n",
       "      <td>8</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12701</th>\n",
       "      <td>военнообязанных</td>\n",
       "      <td>0</td>\n",
       "      <td>15</td>\n",
       "      <td>7</td>\n",
       "      <td>8</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                   Word  Label  Length  Vowels  Consonants  \\\n",
       "11900  Вимм-Билль-Данн Продукты Питания      0      32      10          22   \n",
       "12699                   военнообязанный      0      15       7           8   \n",
       "12700                  военнообязанными      0      16       8           8   \n",
       "12701                   военнообязанных      0      15       7           8   \n",
       "\n",
       "       is_lower  Double  \n",
       "11900         0       3  \n",
       "12699         1       3  \n",
       "12700         1       3  \n",
       "12701         1       3  "
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train[train['Double'] == 3]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##Cross-validation\n",
    "\n",
    "Будем использовать train_test_split - http://scikit-learn.org/stable/modules/generated/sklearn.cross_validation.train_test_split.html\n",
    "\n",
    "Случайным образом выбираем 10 процентов выборки и проверяемся на ней."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Выберем колонки, которые будем использовать для предсказания"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Length</th>\n",
       "      <th>Vowels</th>\n",
       "      <th>Consonants</th>\n",
       "      <th>is_lower</th>\n",
       "      <th>Double</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>8</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>6</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Length  Vowels  Consonants  is_lower  Double\n",
       "0       8       4           4         0       1\n",
       "1       3       2           1         0       1\n",
       "2       5       3           2         0       1\n",
       "3       5       3           2         0       1\n",
       "4       6       4           2         0       1"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cols = train.columns.drop(['Word', 'Label'])\n",
    "train[cols][:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.cross_validation import train_test_split\n",
    "\n",
    "xtrain, xcv, ytrain, ycv = train_test_split(train[cols], train['Label'], test_size = 0.1,  random_state = 10)\n",
    "\n",
    "xtrain, xcv = map(lambda x: df(x, columns=cols), [xtrain, xcv])\n",
    "ytrain, ycv = map(lambda x: df(x, columns=['Label']), [ytrain, ycv])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "xtrain - выборка для обучения, ytrain - ответы к xtrain"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Length</th>\n",
       "      <th>Vowels</th>\n",
       "      <th>Consonants</th>\n",
       "      <th>is_lower</th>\n",
       "      <th>Double</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>78228</th>\n",
       "      <td>9</td>\n",
       "      <td>4</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49860</th>\n",
       "      <td>8</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>87200</th>\n",
       "      <td>7</td>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Length  Vowels  Consonants  is_lower  Double\n",
       "78228       9       4           5         0       0\n",
       "49860       8       4           4         0       0\n",
       "87200       7       2           5         0       0"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "xtrain[:3] "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>78228</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49860</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>87200</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Label\n",
       "78228      0\n",
       "49860      0\n",
       "87200      1"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ytrain[:3]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "xcv - выборка для валидации, ycv - ответы к xcv"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Length</th>\n",
       "      <th>Vowels</th>\n",
       "      <th>Consonants</th>\n",
       "      <th>is_lower</th>\n",
       "      <th>Double</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>42019</th>\n",
       "      <td>7</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4764</th>\n",
       "      <td>5</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42852</th>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Length  Vowels  Consonants  is_lower  Double\n",
       "42019       7       4           3         1       0\n",
       "4764        5       2           3         1       0\n",
       "42852       6       3           3         0       0"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "xcv[:3]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>42019</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4764</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42852</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Label\n",
       "42019      0\n",
       "4764       0\n",
       "42852      0"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ycv[:3]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Параметры логистической регрессии можно посмотреть здесь http://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LogisticRegression.html#sklearn.linear_model.LogisticRegression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.linear_model import LogisticRegression\n",
    "\n",
    "lr = LogisticRegression()\n",
    "lr.fit(xtrain, ytrain)#обучаем модель\n",
    "predictionLR = lr.predict_proba(xcv)#делаем предсказание"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Получили для каждого объекта вероятности его принадлежности к классам"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 0.98142565,  0.01857435],\n",
       "       [ 0.96221468,  0.03778532],\n",
       "       [ 0.79571089,  0.20428911],\n",
       "       [ 0.74172576,  0.25827424],\n",
       "       [ 0.99131361,  0.00868639]])"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "predictionLR[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7924645218185653"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "roc_auc_score(ycv, predictionLR[:,1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Попробуем random forest http://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestClassifier.html"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.80345286115131476"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "\n",
    "rf = RandomForestClassifier(n_estimators = 10)\n",
    "rf.fit(xtrain, ytrain)\n",
    "predictionRF = rf.predict_proba(xcv)\n",
    "\n",
    "roc_auc_score(ycv, predictionRF[:,1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Посмотрим, помогли ли наши странные фичи: запустим те же классификаторы, только без 2 признаков."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.66052015560732447"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "poor_cols = cols.drop(['is_lower', 'Double'])\n",
    "\n",
    "xtrain, xcv, ytrain, ycv = train_test_split(train[poor_cols], train['Label'], test_size = 0.1,  random_state = 10)\n",
    "\n",
    "xtrain, xcv = map(lambda x: df(x, columns=poor_cols), [xtrain, xcv])\n",
    "ytrain, ycv = map(lambda x: df(x, columns=['Label']), [ytrain, ycv])\n",
    "\n",
    "lr = LogisticRegression()\n",
    "lr.fit(xtrain, ytrain)#обучаем модель\n",
    "prediction = lr.predict_proba(xcv)#делаем предсказание\n",
    "roc_auc_score(ycv, prediction[:,1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.66776802547365344"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rf = RandomForestClassifier(n_estimators = 10)\n",
    "rf.fit(xtrain, ytrain)\n",
    "predictionRF = rf.predict_proba(xcv)\n",
    "\n",
    "roc_auc_score(ycv, predictionRF[:,1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##Делаем сабмит"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Word</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>﻿Аалто</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>ААР</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Аара</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Ааре</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Аарон</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     Word\n",
       "0  ﻿Аалто\n",
       "1     ААР\n",
       "2    Аара\n",
       "3    Ааре\n",
       "4   Аарон"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test = pd.read_csv('data/test.csv', names=['Word'])\n",
    "test['Word'] = test['Word'].apply(lambda word: word.decode('utf-8'))\n",
    "\n",
    "test[:5]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "train и test имеют пересечение - оставим это на совести создателей датасета"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2785"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(set(train['Word']).intersection(set(test['Word'])))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Теперь создадим для теста все те же признаки, что и для трэйна"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Word</th>\n",
       "      <th>Length</th>\n",
       "      <th>Vowels</th>\n",
       "      <th>Consonants</th>\n",
       "      <th>is_lower</th>\n",
       "      <th>Double</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>﻿Аалто</td>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>ААР</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Аара</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Ааре</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Аарон</td>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     Word  Length  Vowels  Consonants  is_lower  Double\n",
       "0  ﻿Аалто       6       3           3         1       1\n",
       "1     ААР       3       2           1         0       1\n",
       "2    Аара       4       3           1         0       1\n",
       "3    Ааре       4       3           1         0       1\n",
       "4   Аарон       5       3           2         0       1"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test['Length'] = test['Word'].apply(lambda word: len(word))\n",
    "test['Vowels'] = test['Word'].apply(lambda word: len([letter for letter in word if letter in vowels]))\n",
    "test['Consonants'] = test['Length'] - test['Vowels']\n",
    "test['is_lower'] = test['Word'].apply(lambda word: int( word[0] >= 'а'.decode('utf-8')))\n",
    "test['Double'] = test['Word'].apply(lambda word: countDoubles(word))\n",
    "\n",
    "test[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "rf = RandomForestClassifier(n_estimators = 10)\n",
    "lr = LogisticRegression()\n",
    "\n",
    "rf.fit(train[cols], train['Label'])\n",
    "submitRF = rf.predict_proba(test[cols])\n",
    "\n",
    "lr.fit(train[cols], train['Label'])\n",
    "submitLR = rf.predict_proba(test[cols])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Смешаем результаты логрегресси и рэндом фореста"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "blended = 0.5*submitRF + 0.5*submitLR"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Приведем к нужному формату"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Id</th>\n",
       "      <th>Answer</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0.023404</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>0.039973</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>0.102719</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Id    Answer\n",
       "0   0  0.023404\n",
       "1   1  0.039973\n",
       "2   2  0.000000\n",
       "3   3  0.000000\n",
       "4   4  0.102719"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "submit = pd.DataFrame()\n",
    "submit['Id'] = xrange(len(test))\n",
    "submit['Answer'] = blended[:,1]\n",
    "submit[:5]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "И остается сохранить в файл"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "submit.to_csv('benchmark.csv', index = False)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}