Airline_Tweets_Sentiment_Prediction.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[nltk_data] Downloading package wordnet to /Users/a0j01no/nltk_data...\n",
      "[nltk_data]   Package wordnet is already up-to-date!\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "from nltk.corpus import stopwords\n",
    "from nltk.tokenize import word_tokenize\n",
    "from nltk.stem import WordNetLemmatizer\n",
    "from nltk.stem.porter import PorterStemmer\n",
    "import seaborn as sns\n",
    "import numpy as np\n",
    "import nltk\n",
    "nltk.download('wordnet')\n",
    "import re"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Reading the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "tweets = pd.read_csv(\"Tweets-1.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['@VirginAmerica Thanks!', 'positive'], dtype=object)"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tweets[[\"text\", \"airline_sentiment\"]].values[14]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['tweet_id', 'airline_sentiment', 'airline_sentiment_confidence',\n",
       "       'negativereason', 'negativereason_confidence', 'airline',\n",
       "       'airline_sentiment_gold', 'name', 'negativereason_gold',\n",
       "       'retweet_count', 'text', 'tweet_coord', 'tweet_created',\n",
       "       'tweet_location', 'user_timezone'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tweets.columns"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Number of Airlines"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['Virgin America', 'United', 'Southwest', 'Delta', 'US Airways',\n",
       "       'American'], dtype=object)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.unique(tweets.loc[:, \"airline\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['negative', 'neutral', 'positive'], dtype='object')"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tweets[tweets[\"airline\"] == \"United\"][\"airline_sentiment\"].value_counts().index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<BarContainer object of 3 artists>"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.bar([\"negative\",\"neutral\",\"positive\"],\n",
    "        tweets[tweets[\"airline\"] == \"United\"][\"airline_sentiment\"].value_counts().values)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Plotting sentiment records for all airlines"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'American')"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 6 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=[10,10])\n",
    "plt.subplot(3,2,1)\n",
    "sns.barplot([\"negative\",\"neutral\",\"positive\"],\n",
    "        tweets[tweets[\"airline\"] == \"United\"][\"airline_sentiment\"].value_counts().values).set_title(\"United\")\n",
    "plt.subplot(3,2,2)\n",
    "sns.barplot([\"negative\",\"neutral\",\"positive\"],\n",
    "        tweets[tweets[\"airline\"] == \"Virgin America\"][\"airline_sentiment\"].value_counts().values).set_title(\"Virgin America\")\n",
    "plt.subplot(3,2,3)\n",
    "sns.barplot([\"negative\",\"neutral\",\"positive\"],\n",
    "        tweets[tweets[\"airline\"] == \"Southwest\"][\"airline_sentiment\"].value_counts().values).set_title(\"Southwest\")\n",
    "plt.subplot(3,2,4)\n",
    "sns.barplot([\"negative\",\"neutral\",\"positive\"],\n",
    "        tweets[tweets[\"airline\"] == \"Delta\"][\"airline_sentiment\"].value_counts().values).set_title(\"Delta\")\n",
    "plt.subplot(3,2,5)\n",
    "sns.barplot([\"negative\",\"neutral\",\"positive\"],\n",
    "        tweets[tweets[\"airline\"] == \"US Airways\"][\"airline_sentiment\"].value_counts().values).set_title(\"US Airways\")\n",
    "plt.subplot(3,2,6)\n",
    "sns.barplot([\"negative\",\"neutral\",\"positive\"],\n",
    "        tweets[tweets[\"airline\"] == \"American\"][\"airline_sentiment\"].value_counts().values).set_title(\"American\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>airline</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>airline_sentiment</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>negative</th>\n",
       "      <td>9178</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>neutral</th>\n",
       "      <td>3099</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>positive</th>\n",
       "      <td>2363</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                   airline\n",
       "airline_sentiment         \n",
       "negative              9178\n",
       "neutral               3099\n",
       "positive              2363"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.pivot_table(tweets, values=\"airline\", index= [\"airline_sentiment\"], aggfunc=len)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th>airline_sentiment</th>\n",
       "      <th>negative</th>\n",
       "      <th>neutral</th>\n",
       "      <th>positive</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>airline</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>American</th>\n",
       "      <td>1960</td>\n",
       "      <td>463</td>\n",
       "      <td>336</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Delta</th>\n",
       "      <td>955</td>\n",
       "      <td>723</td>\n",
       "      <td>544</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Southwest</th>\n",
       "      <td>1186</td>\n",
       "      <td>664</td>\n",
       "      <td>570</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>US Airways</th>\n",
       "      <td>2263</td>\n",
       "      <td>381</td>\n",
       "      <td>269</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>United</th>\n",
       "      <td>2633</td>\n",
       "      <td>697</td>\n",
       "      <td>492</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Virgin America</th>\n",
       "      <td>181</td>\n",
       "      <td>171</td>\n",
       "      <td>152</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "airline_sentiment  negative  neutral  positive\n",
       "airline                                       \n",
       "American               1960      463       336\n",
       "Delta                   955      723       544\n",
       "Southwest              1186      664       570\n",
       "US Airways             2263      381       269\n",
       "United                 2633      697       492\n",
       "Virgin America          181      171       152"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.pivot_table(tweets[[\"airline\", \"airline_sentiment\", ]],\n",
    "               index=[\"airline\"], columns= [\"airline_sentiment\"], aggfunc=len)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Customer Service Issue         2910\n",
       "Late Flight                    1665\n",
       "Can't Tell                     1190\n",
       "Cancelled Flight                847\n",
       "Lost Luggage                    724\n",
       "Bad Flight                      580\n",
       "Flight Booking Problems         529\n",
       "Flight Attendant Complaints     481\n",
       "longlines                       178\n",
       "Damaged Luggage                  74\n",
       "Name: negativereason, dtype: int64"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tweets[\"negativereason\"][pd.notna(tweets[\"negativereason\"])].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Washington, DC    111\n",
       "Boston, MA        103\n",
       "New York, NY       93\n",
       "New York           86\n",
       "NYC                68\n",
       "                 ... \n",
       "Novi, MI            1\n",
       "nyc                 1\n",
       "oh so magical.      1\n",
       "Boston,MA           1\n",
       "college             1\n",
       "Name: tweet_location, Length: 1987, dtype: int64"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tweets[\"tweet_location\"][pd.notna(tweets[\"negativereason\"])].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "tweet_nr=tweets[pd.notna(tweets[\"negativereason\"])]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "tweet_loc_nr = tweet_nr[pd.notna(tweet_nr[\"tweet_location\"])]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Reasons for negative sentiments"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['Late Flight', 'Bad Flight', 'Customer Service Issue',\n",
       "       'Flight Booking Problems', 'Lost Luggage',\n",
       "       'Flight Attendant Complaints', \"Can't Tell\", 'Cancelled Flight',\n",
       "       'Damaged Luggage', 'longlines'], dtype=object)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.unique(tweet_loc_nr.negativereason)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "#tweet_loc_nr[tweet_loc_nr[\"negativereason\"] == \"Late Flight\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th>tweet_location</th>\n",
       "      <th>DC | Jersey City</th>\n",
       "      <th>Los Angeles</th>\n",
       "      <th>Mexico, D.F.</th>\n",
       "      <th>NASHVILLE,TN</th>\n",
       "      <th>NYC</th>\n",
       "      <th>Ole Miss • ATL</th>\n",
       "      <th>#BrownsNation</th>\n",
       "      <th>#ManorvilleInExile</th>\n",
       "      <th>#PureMichigan</th>\n",
       "      <th>#Titletown</th>\n",
       "      <th>...</th>\n",
       "      <th>•New York•</th>\n",
       "      <th>⋆city of lost angels⋆</th>\n",
       "      <th>☀️SoFla☀️</th>\n",
       "      <th>☁️</th>\n",
       "      <th>✅</th>\n",
       "      <th>✈</th>\n",
       "      <th>✈️FL/NJ/NYC</th>\n",
       "      <th>✨</th>\n",
       "      <th>❤</th>\n",
       "      <th>명동서식 37.56638,126.984994</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>negativereason</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Bad Flight</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Can't Tell</th>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Cancelled Flight</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Customer Service Issue</th>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Damaged Luggage</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Flight Attendant Complaints</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Flight Booking Problems</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Late Flight</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>3.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Lost Luggage</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>longlines</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>10 rows × 1987 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "tweet_location                DC | Jersey City   Los Angeles   Mexico, D.F.  \\\n",
       "negativereason                                                                \n",
       "Bad Flight                                 NaN           NaN            NaN   \n",
       "Can't Tell                                 NaN           1.0            NaN   \n",
       "Cancelled Flight                           NaN           NaN            NaN   \n",
       "Customer Service Issue                     1.0           NaN            NaN   \n",
       "Damaged Luggage                            NaN           NaN            NaN   \n",
       "Flight Attendant Complaints                NaN           NaN            NaN   \n",
       "Flight Booking Problems                    NaN           NaN            NaN   \n",
       "Late Flight                                NaN           NaN            3.0   \n",
       "Lost Luggage                               NaN           NaN            NaN   \n",
       "longlines                                  NaN           NaN            NaN   \n",
       "\n",
       "tweet_location                NASHVILLE,TN    NYC   Ole Miss • ATL   \\\n",
       "negativereason                                                        \n",
       "Bad Flight                              NaN   NaN               NaN   \n",
       "Can't Tell                              NaN   NaN               NaN   \n",
       "Cancelled Flight                        NaN   NaN               NaN   \n",
       "Customer Service Issue                  1.0   NaN               1.0   \n",
       "Damaged Luggage                         NaN   NaN               NaN   \n",
       "Flight Attendant Complaints             NaN   NaN               NaN   \n",
       "Flight Booking Problems                 NaN   NaN               NaN   \n",
       "Late Flight                             NaN   1.0               NaN   \n",
       "Lost Luggage                            NaN   NaN               NaN   \n",
       "longlines                               NaN   NaN               NaN   \n",
       "\n",
       "tweet_location               #BrownsNation   #ManorvilleInExile  \\\n",
       "negativereason                                                    \n",
       "Bad Flight                              1.0                 NaN   \n",
       "Can't Tell                              NaN                 NaN   \n",
       "Cancelled Flight                        NaN                 NaN   \n",
       "Customer Service Issue                  NaN                 1.0   \n",
       "Damaged Luggage                         NaN                 NaN   \n",
       "Flight Attendant Complaints             NaN                 NaN   \n",
       "Flight Booking Problems                 NaN                 1.0   \n",
       "Late Flight                             NaN                 NaN   \n",
       "Lost Luggage                            NaN                 NaN   \n",
       "longlines                               NaN                 NaN   \n",
       "\n",
       "tweet_location               #PureMichigan   #Titletown  ...  •New York•  \\\n",
       "negativereason                                           ...               \n",
       "Bad Flight                              NaN         NaN  ...         NaN   \n",
       "Can't Tell                              NaN         NaN  ...         NaN   \n",
       "Cancelled Flight                        NaN         NaN  ...         1.0   \n",
       "Customer Service Issue                  1.0         NaN  ...         NaN   \n",
       "Damaged Luggage                         NaN         NaN  ...         NaN   \n",
       "Flight Attendant Complaints             NaN         NaN  ...         NaN   \n",
       "Flight Booking Problems                 NaN         NaN  ...         NaN   \n",
       "Late Flight                             NaN         NaN  ...         NaN   \n",
       "Lost Luggage                            NaN         1.0  ...         NaN   \n",
       "longlines                               NaN         NaN  ...         NaN   \n",
       "\n",
       "tweet_location               ⋆city of lost angels⋆  ☀️SoFla☀️   ☁️    ✅    ✈  \\\n",
       "negativereason                                                                 \n",
       "Bad Flight                                     NaN        NaN  NaN  NaN  NaN   \n",
       "Can't Tell                                     NaN        NaN  1.0  NaN  1.0   \n",
       "Cancelled Flight                               NaN        NaN  NaN  NaN  NaN   \n",
       "Customer Service Issue                         NaN        NaN  NaN  NaN  NaN   \n",
       "Damaged Luggage                                NaN        NaN  NaN  NaN  NaN   \n",
       "Flight Attendant Complaints                    1.0        NaN  NaN  NaN  NaN   \n",
       "Flight Booking Problems                        NaN        NaN  NaN  1.0  NaN   \n",
       "Late Flight                                    NaN        NaN  NaN  NaN  NaN   \n",
       "Lost Luggage                                   NaN        2.0  3.0  NaN  NaN   \n",
       "longlines                                      1.0        NaN  NaN  NaN  NaN   \n",
       "\n",
       "tweet_location               ✈️FL/NJ/NYC    ✨    ❤  명동서식 37.56638,126.984994  \n",
       "negativereason                                                                \n",
       "Bad Flight                           NaN  NaN  NaN                       NaN  \n",
       "Can't Tell                           NaN  NaN  1.0                       NaN  \n",
       "Cancelled Flight                     NaN  NaN  NaN                       1.0  \n",
       "Customer Service Issue               NaN  NaN  NaN                       NaN  \n",
       "Damaged Luggage                      NaN  NaN  NaN                       NaN  \n",
       "Flight Attendant Complaints          NaN  NaN  NaN                       NaN  \n",
       "Flight Booking Problems              NaN  NaN  NaN                       NaN  \n",
       "Late Flight                          2.0  1.0  1.0                       NaN  \n",
       "Lost Luggage                         NaN  NaN  NaN                       NaN  \n",
       "longlines                            NaN  NaN  NaN                       NaN  \n",
       "\n",
       "[10 rows x 1987 columns]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.pivot_table(tweet_loc_nr[[\"negativereason\", \"tweet_location\"]], \n",
    "               index=[\"negativereason\"], columns=[\"tweet_location\"], aggfunc=len)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## The data seems to be sparse so we cannot drill down any further, so let's use ML techniques to learn more"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "X = list(tweets.text.values)\n",
    "Y = list(tweets.airline_sentiment.values)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Checking if all are marked sentiments\n",
    "len(X) == len(Y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "def preprocess(document, stem=False):\n",
    "    document = document.lower()\n",
    "    \n",
    "    words = word_tokenize(document)\n",
    "    \n",
    "    words = [word for word in words if word not in stopwords.words(\"english\")]    # removing stopwords\n",
    "    \n",
    "    if stem:\n",
    "        words = [PorterStemmer().stem(word) for word in words]                     # Stemming the word\n",
    "    else:\n",
    "        words = [WordNetLemmatizer().lemmatize(word, pos='v') for word in words]   # Lemmetizing the word \n",
    "    \n",
    "    document = \" \".join(words)\n",
    "    \n",
    "    return document"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "all_words = []\n",
    "message_set = []\n",
    "for s,l in zip(X,Y):\n",
    "    #print(s)\n",
    "    temp = s\n",
    "    temp = re.sub(r'@[A-Za-z]+', '', temp)                                # removing words with @ signs\n",
    "    temp = re.sub(r'[^\\x00-\\x7F]+', '', re.sub(r\"http\\S+\", \"\", temp))     # removing emoji's\n",
    "    temp = re.sub(r'[^\\w\\s]', '', temp)                                   # removing punctuations\n",
    "    temp = re.sub(r'[0-9]+', '', temp)                                    # removing numbers\n",
    "    filterd_words = [word for word in preprocess(temp).split() if len(word) > 3]\n",
    "    if len(filterd_words) == 0:                             # removing the messages which have no words in it\n",
    "        continue\n",
    "    message_set.append((filterd_words, l))\n",
    "    all_words.extend(filterd_words)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_word_features(all_words):\n",
    "    wordlist = nltk.FreqDist(all_words)\n",
    "    word_features = wordlist.keys()\n",
    "    return word_features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "9664"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(get_word_features(all_words))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "idx_80 = int(len(message_set)*0.8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "11628"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "idx_80"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Taking 80% data for training and 20% for testing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 283,
   "metadata": {},
   "outputs": [],
   "source": [
    "training_messages = message_set[:idx_80]\n",
    "test_messages = message_set[idx_80:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 281,
   "metadata": {},
   "outputs": [],
   "source": [
    "word_features = get_word_features(all_words)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Making the data usable for nltk Naive Bayes classifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 282,
   "metadata": {},
   "outputs": [],
   "source": [
    "def extract_features(document):\n",
    "    doc_words = set(document)\n",
    "    features = {}\n",
    "    for wf in word_features:\n",
    "        features['contains(%s)' % wf] = (wf in doc_words)\n",
    "    return features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 284,
   "metadata": {},
   "outputs": [],
   "source": [
    "training_set = nltk.apply_features(extract_features, training_messages)\n",
    "testing_set = nltk.apply_features(extract_features, test_messages)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Applying NB"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 286,
   "metadata": {},
   "outputs": [],
   "source": [
    "sentiment_classifier = nltk.NaiveBayesClassifier.train(training_set)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 292,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.8632610939112487\n"
     ]
    }
   ],
   "source": [
    "nltk.classify.accuracy(sentiment_classifier, training_set))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Training accuracy is 86%"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 435,
   "metadata": {},
   "outputs": [],
   "source": [
    "nb_Classifier = nltk.classify.accuracy(sentiment_classifier, testing_set)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 291,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Most Informative Features\n",
      "      contains(passbook) = True           positi : negati =     36.7 : 1.0\n",
      "      contains(favorite) = True           positi : negati =     34.3 : 1.0\n",
      "       contains(helpful) = True           positi : neutra =     27.9 : 1.0\n",
      "         contains(daily) = True           neutra : negati =     26.7 : 1.0\n",
      "     contains(beautiful) = True           positi : negati =     24.8 : 1.0\n",
      "          contains(rock) = True           positi : negati =     23.6 : 1.0\n",
      "         contains(hours) = True           negati : positi =     23.3 : 1.0\n",
      "      contains(discount) = True           neutra : negati =     23.0 : 1.0\n",
      "         contains(kudos) = True           positi : negati =     22.8 : 1.0\n",
      "       contains(awesome) = True           positi : negati =     22.2 : 1.0\n",
      "          contains(fail) = True           negati : positi =     21.3 : 1.0\n",
      "       contains(dragons) = True           neutra : negati =     21.1 : 1.0\n",
      "     contains(fantastic) = True           positi : negati =     20.6 : 1.0\n",
      "        contains(battle) = True           neutra : negati =     20.4 : 1.0\n",
      "        contains(street) = True           neutra : negati =     20.4 : 1.0\n",
      "        contains(online) = True           negati : positi =     20.2 : 1.0\n",
      "         contains(smile) = True           positi : negati =     20.1 : 1.0\n",
      "         contains(great) = True           positi : neutra =     19.5 : 1.0\n",
      "         contains(promo) = True           neutra : negati =     19.3 : 1.0\n",
      "          contains(suck) = True           negati : neutra =     19.2 : 1.0\n",
      "     contains(wonderful) = True           positi : neutra =     18.4 : 1.0\n",
      "  contains(professional) = True           positi : negati =     17.7 : 1.0\n",
      "    contains(incredible) = True           positi : negati =     17.7 : 1.0\n",
      "         contains(amaze) = True           positi : negati =     17.5 : 1.0\n",
      "          contains(wall) = True           neutra : negati =     16.9 : 1.0\n",
      "    contains(republican) = True           neutra : negati =     15.6 : 1.0\n",
      "     contains(waterbury) = True           neutra : negati =     15.6 : 1.0\n",
      "    contains(commercial) = True           positi : negati =     15.4 : 1.0\n",
      "     contains(excellent) = True           positi : negati =     15.4 : 1.0\n",
      "       contains(deserve) = True           positi : neutra =     15.0 : 1.0\n",
      "        contains(rebook) = True           negati : positi =     14.9 : 1.0\n",
      "      contains(terrible) = True           negati : positi =     14.2 : 1.0\n",
      "          contains(half) = True           negati : neutra =     14.1 : 1.0\n",
      "          contains(dmed) = True           neutra : negati =     13.8 : 1.0\n",
      "          contains(hold) = True           negati : positi =     13.7 : 1.0\n",
      "          contains(wont) = True           negati : positi =     13.6 : 1.0\n",
      "        contains(agents) = True           negati : neutra =     13.6 : 1.0\n",
      "         contains(flyfi) = True           positi : negati =     13.5 : 1.0\n",
      "          contains(mint) = True           positi : negati =     13.5 : 1.0\n",
      "        contains(strand) = True           negati : positi =     13.4 : 1.0\n",
      "          contains(haha) = True           positi : negati =     13.0 : 1.0\n",
      "      contains(fabulous) = True           positi : negati =     13.0 : 1.0\n",
      "        contains(nicely) = True           positi : negati =     13.0 : 1.0\n",
      "        contains(tarmac) = True           negati : positi =     12.9 : 1.0\n",
      "          contains(hang) = True           negati : neutra =     12.8 : 1.0\n",
      "         contains(visit) = True           neutra : negati =     12.7 : 1.0\n",
      "    contains(experience) = True           negati : neutra =     12.6 : 1.0\n",
      "          contains(hour) = True           negati : neutra =     12.6 : 1.0\n",
      "     contains(frustrate) = True           negati : neutra =     12.5 : 1.0\n",
      "         contains(three) = True           negati : neutra =     12.5 : 1.0\n",
      "          contains(mins) = True           negati : positi =     12.3 : 1.0\n",
      "       contains(appease) = True           neutra : negati =     12.2 : 1.0\n",
      "contains(destinationdragons) = True           neutra : negati =     12.1 : 1.0\n",
      "         contains(sweet) = True           positi : negati =     12.1 : 1.0\n",
      "      contains(thankyou) = True           positi : negati =     12.1 : 1.0\n",
      "        contains(safely) = True           positi : negati =     12.1 : 1.0\n",
      "         contains(among) = True           neutra : negati =     11.9 : 1.0\n",
      "       contains(january) = True           neutra : negati =     11.9 : 1.0\n",
      "         contains(greet) = True           neutra : negati =     11.9 : 1.0\n",
      "       contains(anytime) = True           neutra : negati =     11.9 : 1.0\n",
      "       contains(october) = True           neutra : negati =     11.9 : 1.0\n",
      "         contains(david) = True           neutra : negati =     11.9 : 1.0\n",
      "       contains(nothing) = True           negati : positi =     11.9 : 1.0\n",
      "          contains(view) = True           positi : negati =     11.7 : 1.0\n",
      "       contains(despite) = True           positi : neutra =     11.6 : 1.0\n",
      "         contains(waste) = True           negati : neutra =     11.0 : 1.0\n",
      "         contains(shout) = True           positi : negati =     10.6 : 1.0\n",
      "       contains(efforts) = True           positi : negati =     10.6 : 1.0\n",
      "          contains(wing) = True           positi : negati =     10.6 : 1.0\n",
      "        contains(excite) = True           positi : negati =     10.6 : 1.0\n",
      "       contains(welcome) = True           positi : negati =     10.6 : 1.0\n",
      "   contains(anniversary) = True           positi : negati =     10.6 : 1.0\n",
      "  contains(greatservice) = True           positi : negati =     10.6 : 1.0\n",
      "         contains(notch) = True           positi : negati =     10.6 : 1.0\n",
      "        contains(speedy) = True           positi : negati =     10.6 : 1.0\n",
      "    contains(definitely) = True           positi : negati =     10.6 : 1.0\n",
      "         contains(carry) = True           neutra : positi =     10.6 : 1.0\n",
      "         contains(count) = True           negati : positi =     10.4 : 1.0\n",
      "          contains(poor) = True           negati : positi =     10.2 : 1.0\n",
      "       contains(general) = True           neutra : negati =     10.1 : 1.0\n",
      "   contains(corporation) = True           neutra : negati =     10.1 : 1.0\n",
      "       contains(charity) = True           neutra : negati =     10.1 : 1.0\n",
      "          contains(gain) = True           neutra : negati =     10.1 : 1.0\n",
      "       contains(eastern) = True           neutra : negati =     10.1 : 1.0\n",
      "         contains(allow) = True           neutra : positi =     10.1 : 1.0\n",
      "         contains(thank) = True           positi : negati =     10.1 : 1.0\n",
      "         contains(fleek) = True           neutra : negati =     10.0 : 1.0\n",
      "        contains(answer) = True           negati : neutra =     10.0 : 1.0\n",
      "         contains(spend) = True           negati : neutra =      9.9 : 1.0\n",
      "          contains(crew) = True           positi : neutra =      9.9 : 1.0\n",
      "          contains(rude) = True           negati : neutra =      9.7 : 1.0\n",
      "         contains(cheer) = True           positi : negati =      9.6 : 1.0\n",
      "        contains(august) = True           neutra : negati =      9.4 : 1.0\n",
      "     contains(flightled) = True           negati : positi =      9.3 : 1.0\n",
      "         contains(twice) = True           negati : positi =      9.3 : 1.0\n",
      "       contains(impress) = True           positi : neutra =      9.0 : 1.0\n",
      "        contains(prompt) = True           positi : neutra =      9.0 : 1.0\n",
      "        contains(handle) = True           positi : neutra =      9.0 : 1.0\n",
      "         contains(raise) = True           positi : neutra =      9.0 : 1.0\n",
      "       contains(luggage) = True           negati : positi =      8.9 : 1.0\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "print(sentiment_classifier.show_most_informative_features(100))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Let's try other ML algorithms"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 293,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "v = CountVectorizer(analyzer='word')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 329,
   "metadata": {},
   "outputs": [],
   "source": [
    "messages_train = []\n",
    "label_train = []\n",
    "messages_test = []\n",
    "label_test = []\n",
    "for (i,j) in training_messages:\n",
    "    messages_train.append(\" \".join(i))\n",
    "    label_train.append(j)\n",
    "for (x,y) in test_messages:\n",
    "    messages_test.append(\" \".join(x))\n",
    "    label_test.append(y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "#messages_train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 331,
   "metadata": {},
   "outputs": [],
   "source": [
    "messages_train = v.fit_transform(messages_train)\n",
    "messages_test = v.transform( messages_test)\n",
    "\n",
    "label_test_num = []\n",
    "for i in label_test:\n",
    "    if i == 'negative':\n",
    "        label_test_num.append(0)\n",
    "    if i == 'neutral':\n",
    "        label_test_num.append(-1)\n",
    "    if i == 'positive':\n",
    "        label_test_num.append(1)\n",
    "\n",
    "label_train_num = []\n",
    "for i in label_train:\n",
    "    if i == 'negative':\n",
    "        label_train_num.append(0)\n",
    "    if i == 'neutral':\n",
    "        label_train_num.append(-1)\n",
    "    if i == 'positive':\n",
    "        label_train_num.append(1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Applying Random Forest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 428,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "rf = RandomForestClassifier(n_estimators=250)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 429,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',\n",
       "            max_depth=None, max_features='auto', max_leaf_nodes=None,\n",
       "            min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "            min_samples_leaf=1, min_samples_split=2,\n",
       "            min_weight_fraction_leaf=0.0, n_estimators=250, n_jobs=None,\n",
       "            oob_score=False, random_state=None, verbose=0,\n",
       "            warm_start=False)"
      ]
     },
     "execution_count": 429,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rf.fit(messages_train, label_train_num)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 430,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred = rf.predict(messages_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 431,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred_train = rf.predict(messages_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 537,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import accuracy_score, recall_score, confusion_matrix\n",
    "from sklearn.metrics import precision_score, f1_score, roc_auc_score, roc_curve"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 436,
   "metadata": {},
   "outputs": [],
   "source": [
    "rf_classifier = accuracy_score(pred, label_test_num)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 434,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9926900584795322"
      ]
     },
     "execution_count": 434,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy_score(pred_train, label_train_num)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Applying K Nearest Neighbours"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 395,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.tree import DecisionTreeClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 359,
   "metadata": {},
   "outputs": [],
   "source": [
    "knc = KNeighborsClassifier(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 360,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "KNeighborsClassifier(algorithm='auto', leaf_size=30, metric='minkowski',\n",
       "           metric_params=None, n_jobs=None, n_neighbors=3, p=2,\n",
       "           weights='uniform')"
      ]
     },
     "execution_count": 360,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "knc.fit(messages_train, label_train_num)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 361,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred_knc = knc.predict(messages_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 362,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred_knc_train = knc.predict(messages_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 437,
   "metadata": {},
   "outputs": [],
   "source": [
    "knn_classifier = accuracy_score(pred_knc, label_test_num)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 364,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7102683178534571"
      ]
     },
     "execution_count": 364,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy_score(pred_knc_train, label_train_num)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Applying SVM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 391,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "SVC(C=0.5, cache_size=200, class_weight=None, coef0=0.0,\n",
       "  decision_function_shape='ovr', degree=3, gamma='auto', kernel='poly',\n",
       "  max_iter=-1, probability=True, random_state=None, shrinking=True,\n",
       "  tol=0.001, verbose=False)"
      ]
     },
     "execution_count": 391,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "svc = SVC(kernel='poly', C=0.5, probability=True, gamma='auto')\n",
    "svc.fit(messages_train, label_train_num)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 392,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred_svc = svc.predict(messages_test)\n",
    "pred_svc_train = svc.predict(messages_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 438,
   "metadata": {},
   "outputs": [],
   "source": [
    "svc_classifier = accuracy_score(pred_svc, label_test_num)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 394,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.6081871345029239"
      ]
     },
     "execution_count": 394,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy_score(pred_svc_train, label_train_num)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Applying Decision Tree"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 396,
   "metadata": {},
   "outputs": [],
   "source": [
    "dc = DecisionTreeClassifier()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 397,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "DecisionTreeClassifier(class_weight=None, criterion='gini', max_depth=None,\n",
       "            max_features=None, max_leaf_nodes=None,\n",
       "            min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "            min_samples_leaf=1, min_samples_split=2,\n",
       "            min_weight_fraction_leaf=0.0, presort=False, random_state=None,\n",
       "            splitter='best')"
      ]
     },
     "execution_count": 397,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dc.fit(messages_train, label_train_num)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 398,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred_dc = dc.predict(messages_test)\n",
    "pred_dc_train = dc.predict(messages_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 439,
   "metadata": {},
   "outputs": [],
   "source": [
    "dt_classifier = accuracy_score(pred_dc, label_test_num)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 400,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9926900584795322"
      ]
     },
     "execution_count": 400,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy_score(pred_dc_train, label_train_num)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Applying Adaboost Classifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 401,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.ensemble import AdaBoostClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 402,
   "metadata": {},
   "outputs": [],
   "source": [
    "ada = AdaBoostClassifier()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 403,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "AdaBoostClassifier(algorithm='SAMME.R', base_estimator=None,\n",
       "          learning_rate=1.0, n_estimators=50, random_state=None)"
      ]
     },
     "execution_count": 403,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ada.fit(messages_train, label_train_num)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 404,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred_ada = ada.predict(messages_test)\n",
    "pred_ada_train = ada.predict(messages_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 440,
   "metadata": {},
   "outputs": [],
   "source": [
    "adab_classifier = accuracy_score(pred_ada, label_test_num)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 406,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.6931544547643619"
      ]
     },
     "execution_count": 406,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy_score(pred_ada_train, label_train_num)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Let's plot Accuracies for all the models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 462,
   "metadata": {},
   "outputs": [],
   "source": [
    "classifiers = [nb_Classifier, rf_classifier, knn_classifier, svc_classifier, dt_classifier, adab_classifier]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 463,
   "metadata": {},
   "outputs": [],
   "source": [
    "classifiers\n",
    "cls = [\"naiveb\", \"randomforest\", \"knn\", \"svc\", \"DecisnTree\", \"AdaBoost\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 471,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[Text(0.5, 0, 'Alogrithms')]"
      ]
     },
     "execution_count": 471,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "p = sns.barplot(cls,  classifiers)\n",
    "p.set_title(\"Acc of Various Models\")\n",
    "p.set(ylim=(0, 1))\n",
    "p.set(ylabel=\"Accuracies\")\n",
    "p.set(xlabel=\"Alogrithms\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Let's now mark all sentiments apart from negative as 1 and do the training again    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 473,
   "metadata": {},
   "outputs": [],
   "source": [
    "v = CountVectorizer(analyzer='word')\n",
    "\n",
    "messages_train = []\n",
    "label_train = []\n",
    "messages_test = []\n",
    "label_test = []\n",
    "for (i,j) in training_messages:\n",
    "    messages_train.append(\" \".join(i))\n",
    "    label_train.append(j)\n",
    "for (x,y) in test_messages:\n",
    "    messages_test.append(\" \".join(x))\n",
    "    label_test.append(y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 474,
   "metadata": {},
   "outputs": [],
   "source": [
    "messages_train = v.fit_transform(messages_train)\n",
    "messages_test = v.transform( messages_test)\n",
    "\n",
    "label_test_num = []\n",
    "for i in label_test:\n",
    "    if i == 'negative':\n",
    "        label_test_num.append(0)\n",
    "    else:\n",
    "        label_test_num.append(1)\n",
    "\n",
    "label_train_num = []\n",
    "for i in label_train:\n",
    "    if i == 'negative':\n",
    "        label_train_num.append(0)\n",
    "    else:\n",
    "        label_train_num.append(1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Let's try apply Random Forest with various Estimators size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "estimators = [50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100]\n",
    "rf_accurarices = []\n",
    "rf_recall = []\n",
    "rf_precision = []\n",
    "rf_f1 = []\n",
    "rf_roc_auc_score = []\n",
    "for est in estimators:\n",
    "    rf = RandomForestClassifier(n_estimators=est, max_depth=)\n",
    "    rf.fit(messages_train, label_train_num)\n",
    "    pred = rf.predict(messages_test)\n",
    "    rf_accurarices.append(accuracy_score(pred, label_test_num))\n",
    "    rf_recall.append(recall_score(pred, label_test_num))\n",
    "    rf_precision.append(precision_score(pred, label_test_num))\n",
    "    rf_f1.append(f1_score(pred, label_test_num))\n",
    "    rf_roc_auc_score.append(roc_auc_score(pred, label_test_num))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 551,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(0.65, 0.85)]"
      ]
     },
     "execution_count": 551,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12,7))\n",
    "p = sns.scatterplot(estimators, rf_accurarices, color='red', label=\"Accuracies\")\n",
    "p = sns.scatterplot(estimators,rf_precision, color='green', label=\"Precision\")\n",
    "p = sns.scatterplot(estimators,rf_recall, color='blue', label=\"Recall\")\n",
    "p = sns.scatterplot(estimators,rf_f1, color='brown', label=\"F1 Score\")\n",
    "p = sns.scatterplot(estimators,rf_roc_auc_score, color='black', label=\"ROC AUC Score\")\n",
    "p.set_title(\"RF Accuracies for various Estimators\")\n",
    "p.set(ylabel=\"Accuracies\")\n",
    "p.set(xlabel=\"No. of Estimators\")\n",
    "p.set(ylim=(0.65, 0.85))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## From above we can see we get good accuracy, precision and recall at estimator 100"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 506,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[1789,  254],\n",
       "       [ 291,  573]])"
      ]
     },
     "execution_count": 506,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "confusion_matrix(pred, label_test_num)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 552,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7770593600172757"
      ]
     },
     "execution_count": 552,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "roc_auc_score(pred, label_test_num)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Also the AUC is 0.78 at estimator 100"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Now let's experiment with Random Forest a more with various depths and estimators"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 561,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GridSearchCV(cv=5, error_score='raise-deprecating',\n",
       "       estimator=RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',\n",
       "            max_depth=None, max_features='auto', max_leaf_nodes=None,\n",
       "            min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "            min_samples_leaf=1, min_samples_split=2,\n",
       "            min_weight_fraction_leaf=0.0, n_estimators='warn', n_jobs=None,\n",
       "            oob_score=False, random_state=None, verbose=0,\n",
       "            warm_start=False),\n",
       "       fit_params=None, iid='warn', n_jobs=None,\n",
       "       param_grid={'max_depth': range(5, 25, 5), 'n_estimators': range(50, 101, 5)},\n",
       "       pre_dispatch='2*n_jobs', refit='AUC', return_train_score='warn',\n",
       "       scoring='accuracy', verbose=0)"
      ]
     },
     "execution_count": 561,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.metrics import make_scorer\n",
    "scoring = {'AUC': 'roc_auc', 'Accuracy': make_scorer(accuracy_score)}\n",
    "parameters = {\"max_depth\": range(5, 25, 5), \"n_estimators\" : range(50, 101, 5)}\n",
    "n_folds = 5\n",
    "rf = RandomForestClassifier()\n",
    "rf = GridSearchCV(estimator=rf, param_grid=parameters, cv=n_folds, scoring=\"accuracy\", refit='AUC')\n",
    "rf.fit(messages_train, label_train_num)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 577,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>mean_fit_time</th>\n",
       "      <th>std_fit_time</th>\n",
       "      <th>mean_score_time</th>\n",
       "      <th>std_score_time</th>\n",
       "      <th>param_max_depth</th>\n",
       "      <th>param_n_estimators</th>\n",
       "      <th>params</th>\n",
       "      <th>split0_test_score</th>\n",
       "      <th>split1_test_score</th>\n",
       "      <th>split2_test_score</th>\n",
       "      <th>...</th>\n",
       "      <th>mean_test_score</th>\n",
       "      <th>std_test_score</th>\n",
       "      <th>rank_test_score</th>\n",
       "      <th>split0_train_score</th>\n",
       "      <th>split1_train_score</th>\n",
       "      <th>split2_train_score</th>\n",
       "      <th>split3_train_score</th>\n",
       "      <th>split4_train_score</th>\n",
       "      <th>mean_train_score</th>\n",
       "      <th>std_train_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.297715</td>\n",
       "      <td>0.022366</td>\n",
       "      <td>0.010498</td>\n",
       "      <td>0.000829</td>\n",
       "      <td>5</td>\n",
       "      <td>50</td>\n",
       "      <td>{'max_depth': 5, 'n_estimators': 50}</td>\n",
       "      <td>0.608079</td>\n",
       "      <td>0.60834</td>\n",
       "      <td>0.596989</td>\n",
       "      <td>...</td>\n",
       "      <td>0.605951</td>\n",
       "      <td>0.004481</td>\n",
       "      <td>44</td>\n",
       "      <td>0.608214</td>\n",
       "      <td>0.608149</td>\n",
       "      <td>0.620015</td>\n",
       "      <td>0.608191</td>\n",
       "      <td>0.608513</td>\n",
       "      <td>0.610616</td>\n",
       "      <td>0.004701</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.546448</td>\n",
       "      <td>0.157162</td>\n",
       "      <td>0.021430</td>\n",
       "      <td>0.004979</td>\n",
       "      <td>5</td>\n",
       "      <td>55</td>\n",
       "      <td>{'max_depth': 5, 'n_estimators': 55}</td>\n",
       "      <td>0.608509</td>\n",
       "      <td>0.60834</td>\n",
       "      <td>0.608172</td>\n",
       "      <td>...</td>\n",
       "      <td>0.608273</td>\n",
       "      <td>0.000135</td>\n",
       "      <td>33</td>\n",
       "      <td>0.608214</td>\n",
       "      <td>0.608149</td>\n",
       "      <td>0.608191</td>\n",
       "      <td>0.608191</td>\n",
       "      <td>0.608191</td>\n",
       "      <td>0.608187</td>\n",
       "      <td>0.000021</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.683755</td>\n",
       "      <td>0.007656</td>\n",
       "      <td>0.024099</td>\n",
       "      <td>0.000311</td>\n",
       "      <td>5</td>\n",
       "      <td>60</td>\n",
       "      <td>{'max_depth': 5, 'n_estimators': 60}</td>\n",
       "      <td>0.608079</td>\n",
       "      <td>0.60877</td>\n",
       "      <td>0.608172</td>\n",
       "      <td>...</td>\n",
       "      <td>0.608273</td>\n",
       "      <td>0.000251</td>\n",
       "      <td>33</td>\n",
       "      <td>0.608537</td>\n",
       "      <td>0.608149</td>\n",
       "      <td>0.608191</td>\n",
       "      <td>0.608191</td>\n",
       "      <td>0.608621</td>\n",
       "      <td>0.608338</td>\n",
       "      <td>0.000199</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.746982</td>\n",
       "      <td>0.010506</td>\n",
       "      <td>0.026380</td>\n",
       "      <td>0.000790</td>\n",
       "      <td>5</td>\n",
       "      <td>65</td>\n",
       "      <td>{'max_depth': 5, 'n_estimators': 65}</td>\n",
       "      <td>0.608079</td>\n",
       "      <td>0.60834</td>\n",
       "      <td>0.608172</td>\n",
       "      <td>...</td>\n",
       "      <td>0.608187</td>\n",
       "      <td>0.000085</td>\n",
       "      <td>37</td>\n",
       "      <td>0.608214</td>\n",
       "      <td>0.608149</td>\n",
       "      <td>0.608191</td>\n",
       "      <td>0.608191</td>\n",
       "      <td>0.608191</td>\n",
       "      <td>0.608187</td>\n",
       "      <td>0.000021</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.794421</td>\n",
       "      <td>0.007573</td>\n",
       "      <td>0.027796</td>\n",
       "      <td>0.000251</td>\n",
       "      <td>5</td>\n",
       "      <td>70</td>\n",
       "      <td>{'max_depth': 5, 'n_estimators': 70}</td>\n",
       "      <td>0.608079</td>\n",
       "      <td>0.60834</td>\n",
       "      <td>0.608172</td>\n",
       "      <td>...</td>\n",
       "      <td>0.608273</td>\n",
       "      <td>0.000185</td>\n",
       "      <td>33</td>\n",
       "      <td>0.608214</td>\n",
       "      <td>0.608149</td>\n",
       "      <td>0.608406</td>\n",
       "      <td>0.608191</td>\n",
       "      <td>0.608298</td>\n",
       "      <td>0.608252</td>\n",
       "      <td>0.000091</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 22 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   mean_fit_time  std_fit_time  mean_score_time  std_score_time  \\\n",
       "0       0.297715      0.022366         0.010498        0.000829   \n",
       "1       0.546448      0.157162         0.021430        0.004979   \n",
       "2       0.683755      0.007656         0.024099        0.000311   \n",
       "3       0.746982      0.010506         0.026380        0.000790   \n",
       "4       0.794421      0.007573         0.027796        0.000251   \n",
       "\n",
       "  param_max_depth param_n_estimators                                params  \\\n",
       "0               5                 50  {'max_depth': 5, 'n_estimators': 50}   \n",
       "1               5                 55  {'max_depth': 5, 'n_estimators': 55}   \n",
       "2               5                 60  {'max_depth': 5, 'n_estimators': 60}   \n",
       "3               5                 65  {'max_depth': 5, 'n_estimators': 65}   \n",
       "4               5                 70  {'max_depth': 5, 'n_estimators': 70}   \n",
       "\n",
       "   split0_test_score  split1_test_score  split2_test_score  ...  \\\n",
       "0           0.608079            0.60834           0.596989  ...   \n",
       "1           0.608509            0.60834           0.608172  ...   \n",
       "2           0.608079            0.60877           0.608172  ...   \n",
       "3           0.608079            0.60834           0.608172  ...   \n",
       "4           0.608079            0.60834           0.608172  ...   \n",
       "\n",
       "   mean_test_score  std_test_score  rank_test_score  split0_train_score  \\\n",
       "0         0.605951        0.004481               44            0.608214   \n",
       "1         0.608273        0.000135               33            0.608214   \n",
       "2         0.608273        0.000251               33            0.608537   \n",
       "3         0.608187        0.000085               37            0.608214   \n",
       "4         0.608273        0.000185               33            0.608214   \n",
       "\n",
       "   split1_train_score  split2_train_score  split3_train_score  \\\n",
       "0            0.608149            0.620015            0.608191   \n",
       "1            0.608149            0.608191            0.608191   \n",
       "2            0.608149            0.608191            0.608191   \n",
       "3            0.608149            0.608191            0.608191   \n",
       "4            0.608149            0.608406            0.608191   \n",
       "\n",
       "   split4_train_score  mean_train_score  std_train_score  \n",
       "0            0.608513          0.610616         0.004701  \n",
       "1            0.608191          0.608187         0.000021  \n",
       "2            0.608621          0.608338         0.000199  \n",
       "3            0.608191          0.608187         0.000021  \n",
       "4            0.608298          0.608252         0.000091  \n",
       "\n",
       "[5 rows x 22 columns]"
      ]
     },
     "execution_count": 577,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "scores = rf.cv_results_\n",
    "pd.DataFrame(rf.cv_results_).head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 581,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['mean_fit_time', 'std_fit_time', 'mean_score_time', 'std_score_time', 'param_max_depth', 'param_n_estimators', 'params', 'split0_test_score', 'split1_test_score', 'split2_test_score', 'split3_test_score', 'split4_test_score', 'mean_test_score', 'std_test_score', 'rank_test_score', 'split0_train_score', 'split1_train_score', 'split2_train_score', 'split3_train_score', 'split4_train_score', 'mean_train_score', 'std_train_score'])"
      ]
     },
     "execution_count": 581,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "scores.keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 582,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x15f5c4350>"
      ]
     },
     "execution_count": 582,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(scores[\"mean_train_score\"], label=\"train accuracy\")\n",
    "plt.plot(scores[\"mean_test_score\"], label=\"test accuracy\")\n",
    "plt.legend()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}