4. Logistic regression, Linear-SVM, XGBoost with hyperparameter tuning+sim fs+Adv fs+Word2Vec Features.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h1><p style=\"color: blue\">Performing hyperparameter tuning of Logistic regression, Linear-SVM, XGBoost with Word2Vec Features</p></h1>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "        <script type=\"text/javascript\">\n",
       "        window.PlotlyConfig = {MathJaxConfig: 'local'};\n",
       "        if (window.MathJax) {MathJax.Hub.Config({SVG: {font: \"STIX-Web\"}});}\n",
       "        if (typeof require !== 'undefined') {\n",
       "        require.undef(\"plotly\");\n",
       "        requirejs.config({\n",
       "            paths: {\n",
       "                'plotly': ['https://cdn.plot.ly/plotly-latest.min']\n",
       "            }\n",
       "        });\n",
       "        require(['plotly'], function(Plotly) {\n",
       "            window._Plotly = Plotly;\n",
       "        });\n",
       "        }\n",
       "        </script>\n",
       "        "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "from subprocess import check_output\n",
    "%matplotlib inline\n",
    "import plotly.offline as py\n",
    "py.init_notebook_mode(connected=True)\n",
    "import plotly.graph_objs as go\n",
    "import plotly.tools as tls\n",
    "import os\n",
    "import gc\n",
    "import re\n",
    "from os import path\n",
    "\n",
    "#import cPickle\n",
    "import _pickle as cPickle\n",
    "import gensim\n",
    "from nltk.corpus import stopwords\n",
    "from tqdm import tqdm\n",
    "from scipy.stats import skew, kurtosis\n",
    "from scipy.spatial.distance import cosine, cityblock, jaccard, canberra, euclidean, minkowski, braycurtis\n",
    "from nltk import word_tokenize\n",
    "\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "\n",
    "from sklearn.manifold import TSNE\n",
    "from sklearn.preprocessing import normalize\n",
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "from sklearn.feature_extraction.text import TfidfVectorizer\n",
    "from sklearn.decomposition import TruncatedSVD\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from sklearn.metrics.classification import accuracy_score, log_loss\n",
    "from sklearn.multiclass import OneVsRestClassifier\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.model_selection import StratifiedKFold \n",
    "from sklearn.calibration import CalibratedClassifierCV\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.metrics import normalized_mutual_info_score\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.model_selection import cross_val_score\n",
    "from sklearn.linear_model import SGDClassifier\n",
    "from sklearn import model_selection\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.metrics import precision_recall_curve, auc, roc_curve\n",
    "from mlxtend.classifier import StackingClassifier\n",
    "\n",
    "from collections import Counter, defaultdict\n",
    "from scipy.sparse import hstack\n",
    "\n",
    "from sqlalchemy import create_engine # database connection\n",
    "\n",
    "import xgboost as xgb\n",
    "\n",
    "from xgboost import XGBClassifier\n",
    "from sklearn.model_selection import RandomizedSearchCV,StratifiedKFold\n",
    "\n",
    "from sklearn.utils.testing import ignore_warnings\n",
    "from sklearn.exceptions import ConvergenceWarning\n",
    "\n",
    "import warnings\n",
    "with warnings.catch_warnings():\n",
    "    warnings.filterwarnings(\"ignore\", category=ConvergenceWarning)\n",
    "\n",
    "#import warnings\n",
    "#with warnings.catch_warnings():\n",
    "#    warnings.filterwarnings(\"ignore\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h3> 5. Loading Data Sheet and getting ready for apply Word2Vec Feature</h3>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 199,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load Basic Feature and advance feature data set\n",
    "df_basic_feature = pd.read_csv(\"df_fe_without_preprocessing_train.csv\",encoding='latin-1')\n",
    "df_advance_feature = pd.read_csv(\"nlp_features_train.csv\",encoding='latin-1')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<p style=\"color: blue\">Now drop 'qid1', 'qid2','question1','question2','is_duplicate' columns and add both the dataframe</p>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 200,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Columns dropped from basic feature dataframe \n",
    "df_basic_feature = df_basic_feature.drop(['qid1','qid2'],axis=1)\n",
    "\n",
    "# Columns dropped from advance feature dataframe \n",
    "df_advance_feature = df_advance_feature.drop(['qid1','qid2','question1','question2','is_duplicate'],axis=1)\n",
    "\n",
    "# Lets add both the truncated dataframe into one dataframe\n",
    "data  = df_basic_feature.merge(df_advance_feature, on='id',how='left')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 201,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "            id                         question1  \\\n",
      "105780  105780    How can I develop android app?   \n",
      "201841  201841  How can I create an Android app?   \n",
      "363362  363362                               NaN   \n",
      "\n",
      "                                                question2  is_duplicate  \\\n",
      "105780                                                NaN             0   \n",
      "201841                                                NaN             0   \n",
      "363362  My Chinese name is Haichao Yu. What English na...             0   \n",
      "\n",
      "        freq_qid1  freq_qid2  q1len  q2len  q1_n_words  q2_n_words  ...  \\\n",
      "105780          2          2     30      0           6           1  ...   \n",
      "201841          1          2     32      0           7           1  ...   \n",
      "363362          1          1      0    123           1          21  ...   \n",
      "\n",
      "        ctc_max  last_word_eq  first_word_eq  abs_len_diff  mean_len  \\\n",
      "105780      0.0           0.0            0.0           0.0       0.0   \n",
      "201841      0.0           0.0            0.0           0.0       0.0   \n",
      "363362      0.0           0.0            0.0           0.0       0.0   \n",
      "\n",
      "        token_set_ratio  token_sort_ratio  fuzz_ratio  fuzz_partial_ratio  \\\n",
      "105780                0                 0           0                   0   \n",
      "201841                0                 0           0                   0   \n",
      "363362                0                 0           0                   0   \n",
      "\n",
      "        longest_substr_ratio  \n",
      "105780                   0.0  \n",
      "201841                   0.0  \n",
      "363362                   0.0  \n",
      "\n",
      "[3 rows x 30 columns]\n"
     ]
    }
   ],
   "source": [
    "nan_rows = data[data.isnull().any(1)]\n",
    "print (nan_rows)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<p style=\"color: blue\">Found 3 such rows, we will remove these rows from the original dataset itself</p>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 202,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = data[data['question1'].notnull()]\n",
    "data = data[data['question2'].notnull()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 203,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Empty DataFrame\n",
      "Columns: [id, question1, question2, is_duplicate, freq_qid1, freq_qid2, q1len, q2len, q1_n_words, q2_n_words, word_Common, word_Total, word_share, freq_q1+q2, freq_q1-q2, cwc_min, cwc_max, csc_min, csc_max, ctc_min, ctc_max, last_word_eq, first_word_eq, abs_len_diff, mean_len, token_set_ratio, token_sort_ratio, fuzz_ratio, fuzz_partial_ratio, longest_substr_ratio]\n",
      "Index: []\n",
      "\n",
      "[0 rows x 30 columns]\n"
     ]
    }
   ],
   "source": [
    "nan_rows = data[data.isnull().any(1)]\n",
    "print (nan_rows)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 204,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 404287 entries, 0 to 404289\n",
      "Data columns (total 30 columns):\n",
      " #   Column                Non-Null Count   Dtype  \n",
      "---  ------                --------------   -----  \n",
      " 0   id                    404287 non-null  int64  \n",
      " 1   question1             404287 non-null  object \n",
      " 2   question2             404287 non-null  object \n",
      " 3   is_duplicate          404287 non-null  int64  \n",
      " 4   freq_qid1             404287 non-null  int64  \n",
      " 5   freq_qid2             404287 non-null  int64  \n",
      " 6   q1len                 404287 non-null  int64  \n",
      " 7   q2len                 404287 non-null  int64  \n",
      " 8   q1_n_words            404287 non-null  int64  \n",
      " 9   q2_n_words            404287 non-null  int64  \n",
      " 10  word_Common           404287 non-null  float64\n",
      " 11  word_Total            404287 non-null  float64\n",
      " 12  word_share            404287 non-null  float64\n",
      " 13  freq_q1+q2            404287 non-null  int64  \n",
      " 14  freq_q1-q2            404287 non-null  int64  \n",
      " 15  cwc_min               404287 non-null  float64\n",
      " 16  cwc_max               404287 non-null  float64\n",
      " 17  csc_min               404287 non-null  float64\n",
      " 18  csc_max               404287 non-null  float64\n",
      " 19  ctc_min               404287 non-null  float64\n",
      " 20  ctc_max               404287 non-null  float64\n",
      " 21  last_word_eq          404287 non-null  float64\n",
      " 22  first_word_eq         404287 non-null  float64\n",
      " 23  abs_len_diff          404287 non-null  float64\n",
      " 24  mean_len              404287 non-null  float64\n",
      " 25  token_set_ratio       404287 non-null  int64  \n",
      " 26  token_sort_ratio      404287 non-null  int64  \n",
      " 27  fuzz_ratio            404287 non-null  int64  \n",
      " 28  fuzz_partial_ratio    404287 non-null  int64  \n",
      " 29  longest_substr_ratio  404287 non-null  float64\n",
      "dtypes: float64(14), int64(14), object(2)\n",
      "memory usage: 95.6+ MB\n"
     ]
    }
   ],
   "source": [
    "data.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 205,
   "metadata": {
    "scrolled": true
   },
   "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>id</th>\n",
       "      <th>question1</th>\n",
       "      <th>question2</th>\n",
       "      <th>is_duplicate</th>\n",
       "      <th>freq_qid1</th>\n",
       "      <th>freq_qid2</th>\n",
       "      <th>q1len</th>\n",
       "      <th>q2len</th>\n",
       "      <th>q1_n_words</th>\n",
       "      <th>q2_n_words</th>\n",
       "      <th>...</th>\n",
       "      <th>ctc_max</th>\n",
       "      <th>last_word_eq</th>\n",
       "      <th>first_word_eq</th>\n",
       "      <th>abs_len_diff</th>\n",
       "      <th>mean_len</th>\n",
       "      <th>token_set_ratio</th>\n",
       "      <th>token_sort_ratio</th>\n",
       "      <th>fuzz_ratio</th>\n",
       "      <th>fuzz_partial_ratio</th>\n",
       "      <th>longest_substr_ratio</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>What is the step by step guide to invest in sh...</td>\n",
       "      <td>What is the step by step guide to invest in sh...</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>66</td>\n",
       "      <td>57</td>\n",
       "      <td>14</td>\n",
       "      <td>12</td>\n",
       "      <td>...</td>\n",
       "      <td>0.785709</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>13.0</td>\n",
       "      <td>100</td>\n",
       "      <td>93</td>\n",
       "      <td>93</td>\n",
       "      <td>100</td>\n",
       "      <td>0.982759</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>What is the story of Kohinoor (Koh-i-Noor) Dia...</td>\n",
       "      <td>What would happen if the Indian government sto...</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>51</td>\n",
       "      <td>88</td>\n",
       "      <td>8</td>\n",
       "      <td>13</td>\n",
       "      <td>...</td>\n",
       "      <td>0.466664</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>12.5</td>\n",
       "      <td>86</td>\n",
       "      <td>63</td>\n",
       "      <td>66</td>\n",
       "      <td>75</td>\n",
       "      <td>0.596154</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>2 rows × 30 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   id                                          question1  \\\n",
       "0   0  What is the step by step guide to invest in sh...   \n",
       "1   1  What is the story of Kohinoor (Koh-i-Noor) Dia...   \n",
       "\n",
       "                                           question2  is_duplicate  freq_qid1  \\\n",
       "0  What is the step by step guide to invest in sh...             0          1   \n",
       "1  What would happen if the Indian government sto...             0          4   \n",
       "\n",
       "   freq_qid2  q1len  q2len  q1_n_words  q2_n_words  ...   ctc_max  \\\n",
       "0          1     66     57          14          12  ...  0.785709   \n",
       "1          1     51     88           8          13  ...  0.466664   \n",
       "\n",
       "   last_word_eq  first_word_eq  abs_len_diff  mean_len  token_set_ratio  \\\n",
       "0           0.0            1.0           2.0      13.0              100   \n",
       "1           0.0            1.0           5.0      12.5               86   \n",
       "\n",
       "   token_sort_ratio  fuzz_ratio  fuzz_partial_ratio  longest_substr_ratio  \n",
       "0                93          93                 100              0.982759  \n",
       "1                63          66                  75              0.596154  \n",
       "\n",
       "[2 rows x 30 columns]"
      ]
     },
     "execution_count": 205,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 206,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "404287"
      ]
     },
     "execution_count": 206,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.shape[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 207,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "202144"
      ]
     },
     "execution_count": 207,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data= data.sample(frac = 0.5) \n",
    "data.shape[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h3>5.1 let's start to make Word2Vec Features and save in dataframe \"data\"</h3>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 209,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "202144it [00:45, 4465.74it/s]\n",
      "202144it [00:39, 5089.00it/s]\n"
     ]
    }
   ],
   "source": [
    "\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\", category=RuntimeWarning) \n",
    "\n",
    "stop_words = stopwords.words('english')\n",
    "\n",
    "\n",
    "def wmd(s1, s2):\n",
    "    s1 = str(s1).lower().split()\n",
    "    s2 = str(s2).lower().split()\n",
    "    stop_words = stopwords.words('english')\n",
    "    s1 = [w for w in s1 if w not in stop_words]\n",
    "    s2 = [w for w in s2 if w not in stop_words]\n",
    "    return model.wmdistance(s1, s2)\n",
    "\n",
    "\n",
    "def norm_wmd(s1, s2):\n",
    "    s1 = str(s1).lower().split()\n",
    "    s2 = str(s2).lower().split()\n",
    "    stop_words = stopwords.words('english')\n",
    "    s1 = [w for w in s1 if w not in stop_words]\n",
    "    s2 = [w for w in s2 if w not in stop_words]\n",
    "    return norm_model.wmdistance(s1, s2)\n",
    "\n",
    "\n",
    "def sent2vec(s):\n",
    "    words = str(s).lower()\n",
    "    words = word_tokenize(words)\n",
    "    words = [w for w in words if not w in stop_words]\n",
    "    words = [w for w in words if w.isalpha()]\n",
    "    M = []\n",
    "    for w in words:\n",
    "        try:\n",
    "            M.append(model[w])\n",
    "        except:\n",
    "            continue\n",
    "    M = np.array(M)\n",
    "    v = M.sum(axis=0)\n",
    "    return v / np.sqrt((v ** 2).sum())\n",
    "\n",
    "\n",
    "\n",
    "model = gensim.models.KeyedVectors.load_word2vec_format('Data/GoogleNews-vectors-negative300.bin.gz', binary=True)\n",
    "#model = gensim.models.KeyedVectors.load_word2vec_format(wv, binary=True)\n",
    "data['wmd'] = data.apply(lambda x: wmd(x['question1'], x['question2']), axis=1)\n",
    "\n",
    "\n",
    "norm_model = gensim.models.KeyedVectors.load_word2vec_format('Data/GoogleNews-vectors-negative300.bin.gz', binary=True)\n",
    "#norm_model = gensim.models.KeyedVectors.load_word2vec_format(wv, binary=True)\n",
    "norm_model.init_sims(replace=True)\n",
    "data['norm_wmd'] = data.apply(lambda x: norm_wmd(x['question1'], x['question2']), axis=1)\n",
    "\n",
    "\n",
    "question1_vectors = np.zeros((data.shape[0], 300))\n",
    "error_count = 0\n",
    "for i, q in tqdm(enumerate(data.question1.values)):\n",
    "    question1_vectors[i, :] = sent2vec(q)\n",
    "\n",
    "question2_vectors  = np.zeros((data.shape[0], 300))\n",
    "for i, q in tqdm(enumerate(data.question2.values)):\n",
    "    question2_vectors[i, :] = sent2vec(q)\n",
    "\n",
    "\n",
    "data['cosine_distance'] = [cosine(x, y) for (x, y) in zip(np.nan_to_num(question1_vectors),\n",
    "                                                          np.nan_to_num(question2_vectors))]\n",
    "\n",
    "data['cityblock_distance'] = [cityblock(x, y) for (x, y) in zip(np.nan_to_num(question1_vectors),\n",
    "                                                          np.nan_to_num(question2_vectors))]\n",
    "\n",
    "data['jaccard_distance'] = [jaccard(x, y) for (x, y) in zip(np.nan_to_num(question1_vectors),\n",
    "                                                          np.nan_to_num(question2_vectors))]\n",
    "\n",
    "data['canberra_distance'] = [canberra(x, y) for (x, y) in zip(np.nan_to_num(question1_vectors),\n",
    "                                                          np.nan_to_num(question2_vectors))]\n",
    "\n",
    "data['euclidean_distance'] = [euclidean(x, y) for (x, y) in zip(np.nan_to_num(question1_vectors),\n",
    "                                                          np.nan_to_num(question2_vectors))]\n",
    "\n",
    "data['minkowski_distance'] = [minkowski(x, y, 3) for (x, y) in zip(np.nan_to_num(question1_vectors),\n",
    "                                                          np.nan_to_num(question2_vectors))]\n",
    "\n",
    "data['braycurtis_distance'] = [braycurtis(x, y) for (x, y) in zip(np.nan_to_num(question1_vectors),\n",
    "                                                          np.nan_to_num(question2_vectors))]\n",
    "\n",
    "data['skew_q1vec'] = [skew(x) for x in np.nan_to_num(question1_vectors)]\n",
    "data['skew_q2vec'] = [skew(x) for x in np.nan_to_num(question2_vectors)]\n",
    "data['kur_q1vec'] = [kurtosis(x) for x in np.nan_to_num(question1_vectors)]\n",
    "data['kur_q2vec'] = [kurtosis(x) for x in np.nan_to_num(question2_vectors)]\n",
    "\n",
    "cPickle.dump(question1_vectors, open('data/q1_w2v.pkl', 'wb'), -1)\n",
    "cPickle.dump(question2_vectors, open('data/q2_w2v.pkl', 'wb'), -1)\n",
    "\n",
    "\n",
    "data.to_csv('data/quora_features.csv', index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h2> 5.2 Applying ML Models on newly created DATA </h2>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "data= pd.read_csv(\"Data/quora_features.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "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>id</th>\n",
       "      <th>question1</th>\n",
       "      <th>question2</th>\n",
       "      <th>is_duplicate</th>\n",
       "      <th>freq_qid1</th>\n",
       "      <th>freq_qid2</th>\n",
       "      <th>q1len</th>\n",
       "      <th>q2len</th>\n",
       "      <th>q1_n_words</th>\n",
       "      <th>q2_n_words</th>\n",
       "      <th>...</th>\n",
       "      <th>cityblock_distance</th>\n",
       "      <th>jaccard_distance</th>\n",
       "      <th>canberra_distance</th>\n",
       "      <th>euclidean_distance</th>\n",
       "      <th>minkowski_distance</th>\n",
       "      <th>braycurtis_distance</th>\n",
       "      <th>skew_q1vec</th>\n",
       "      <th>skew_q2vec</th>\n",
       "      <th>kur_q1vec</th>\n",
       "      <th>kur_q2vec</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>Why do white women initiate divorce so much?</td>\n",
       "      <td>Why do women initiate the majority of divorces?</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>44</td>\n",
       "      <td>47</td>\n",
       "      <td>8</td>\n",
       "      <td>8</td>\n",
       "      <td>...</td>\n",
       "      <td>8.690568</td>\n",
       "      <td>1.0</td>\n",
       "      <td>125.797689</td>\n",
       "      <td>0.646990</td>\n",
       "      <td>0.294016</td>\n",
       "      <td>0.322897</td>\n",
       "      <td>0.110333</td>\n",
       "      <td>0.253555</td>\n",
       "      <td>-0.472383</td>\n",
       "      <td>0.007384</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>Is there a way on Quora to ask why a specific ...</td>\n",
       "      <td>All my questions on Quora need improvement. Wh...</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>17</td>\n",
       "      <td>87</td>\n",
       "      <td>92</td>\n",
       "      <td>17</td>\n",
       "      <td>18</td>\n",
       "      <td>...</td>\n",
       "      <td>6.874296</td>\n",
       "      <td>1.0</td>\n",
       "      <td>111.968110</td>\n",
       "      <td>0.508915</td>\n",
       "      <td>0.232597</td>\n",
       "      <td>0.255839</td>\n",
       "      <td>0.100403</td>\n",
       "      <td>-0.047450</td>\n",
       "      <td>-0.011347</td>\n",
       "      <td>-0.036685</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>2 rows × 43 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   id                                          question1  \\\n",
       "0   1       Why do white women initiate divorce so much?   \n",
       "1   2  Is there a way on Quora to ask why a specific ...   \n",
       "\n",
       "                                           question2  is_duplicate  freq_qid1  \\\n",
       "0    Why do women initiate the majority of divorces?             1          1   \n",
       "1  All my questions on Quora need improvement. Wh...             1          3   \n",
       "\n",
       "   freq_qid2  q1len  q2len  q1_n_words  q2_n_words  ...  cityblock_distance  \\\n",
       "0          1     44     47           8           8  ...            8.690568   \n",
       "1         17     87     92          17          18  ...            6.874296   \n",
       "\n",
       "   jaccard_distance  canberra_distance  euclidean_distance  \\\n",
       "0               1.0         125.797689            0.646990   \n",
       "1               1.0         111.968110            0.508915   \n",
       "\n",
       "   minkowski_distance  braycurtis_distance  skew_q1vec  skew_q2vec  kur_q1vec  \\\n",
       "0            0.294016             0.322897    0.110333    0.253555  -0.472383   \n",
       "1            0.232597             0.255839    0.100403   -0.047450  -0.011347   \n",
       "\n",
       "   kur_q2vec  \n",
       "0   0.007384  \n",
       "1  -0.036685  \n",
       "\n",
       "[2 rows x 43 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Empty DataFrame\n",
      "Columns: [id, question1, question2, is_duplicate, freq_qid1, freq_qid2, q1len, q2len, q1_n_words, q2_n_words, word_Common, word_Total, word_share, freq_q1+q2, freq_q1-q2, cwc_min, cwc_max, csc_min, csc_max, ctc_min, ctc_max, last_word_eq, first_word_eq, abs_len_diff, mean_len, token_set_ratio, token_sort_ratio, fuzz_ratio, fuzz_partial_ratio, longest_substr_ratio, wmd, norm_wmd, cosine_distance, cityblock_distance, jaccard_distance, canberra_distance, euclidean_distance, minkowski_distance, braycurtis_distance, skew_q1vec, skew_q2vec, kur_q1vec, kur_q2vec]\n",
      "Index: []\n",
      "\n",
      "[0 rows x 43 columns]\n"
     ]
    }
   ],
   "source": [
    "nan_rows = data[data.isnull().any(1)]\n",
    "print (nan_rows)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Drop unnecessary question1 and question2 columns from \"data\" \n",
    "data.drop(['question1','question2'], axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "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>id</th>\n",
       "      <th>is_duplicate</th>\n",
       "      <th>freq_qid1</th>\n",
       "      <th>freq_qid2</th>\n",
       "      <th>q1len</th>\n",
       "      <th>q2len</th>\n",
       "      <th>q1_n_words</th>\n",
       "      <th>q2_n_words</th>\n",
       "      <th>word_Common</th>\n",
       "      <th>word_Total</th>\n",
       "      <th>...</th>\n",
       "      <th>cityblock_distance</th>\n",
       "      <th>jaccard_distance</th>\n",
       "      <th>canberra_distance</th>\n",
       "      <th>euclidean_distance</th>\n",
       "      <th>minkowski_distance</th>\n",
       "      <th>braycurtis_distance</th>\n",
       "      <th>skew_q1vec</th>\n",
       "      <th>skew_q2vec</th>\n",
       "      <th>kur_q1vec</th>\n",
       "      <th>kur_q2vec</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>44</td>\n",
       "      <td>47</td>\n",
       "      <td>8</td>\n",
       "      <td>8</td>\n",
       "      <td>4</td>\n",
       "      <td>16</td>\n",
       "      <td>...</td>\n",
       "      <td>8.690568</td>\n",
       "      <td>1.0</td>\n",
       "      <td>125.797689</td>\n",
       "      <td>0.646990</td>\n",
       "      <td>0.294016</td>\n",
       "      <td>0.322897</td>\n",
       "      <td>0.110333</td>\n",
       "      <td>0.253555</td>\n",
       "      <td>-0.472383</td>\n",
       "      <td>0.007384</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>17</td>\n",
       "      <td>87</td>\n",
       "      <td>92</td>\n",
       "      <td>17</td>\n",
       "      <td>18</td>\n",
       "      <td>8</td>\n",
       "      <td>33</td>\n",
       "      <td>...</td>\n",
       "      <td>6.874296</td>\n",
       "      <td>1.0</td>\n",
       "      <td>111.968110</td>\n",
       "      <td>0.508915</td>\n",
       "      <td>0.232597</td>\n",
       "      <td>0.255839</td>\n",
       "      <td>0.100403</td>\n",
       "      <td>-0.047450</td>\n",
       "      <td>-0.011347</td>\n",
       "      <td>-0.036685</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>2 rows × 41 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   id  is_duplicate  freq_qid1  freq_qid2  q1len  q2len  q1_n_words  \\\n",
       "0   1             1          1          1     44     47           8   \n",
       "1   2             1          3         17     87     92          17   \n",
       "\n",
       "   q2_n_words  word_Common  word_Total  ...  cityblock_distance  \\\n",
       "0           8            4          16  ...            8.690568   \n",
       "1          18            8          33  ...            6.874296   \n",
       "\n",
       "   jaccard_distance  canberra_distance  euclidean_distance  \\\n",
       "0               1.0         125.797689            0.646990   \n",
       "1               1.0         111.968110            0.508915   \n",
       "\n",
       "   minkowski_distance  braycurtis_distance  skew_q1vec  skew_q2vec  kur_q1vec  \\\n",
       "0            0.294016             0.322897    0.110333    0.253555  -0.472383   \n",
       "1            0.232597             0.255839    0.100403   -0.047450  -0.011347   \n",
       "\n",
       "   kur_q2vec  \n",
       "0   0.007384  \n",
       "1  -0.036685  \n",
       "\n",
       "[2 rows x 41 columns]"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "196433"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.shape[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_true = data['is_duplicate']\n",
    "\n",
    "data.drop(['id','is_duplicate'], axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "scrolled": true
   },
   "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>freq_qid1</th>\n",
       "      <th>freq_qid2</th>\n",
       "      <th>q1len</th>\n",
       "      <th>q2len</th>\n",
       "      <th>q1_n_words</th>\n",
       "      <th>q2_n_words</th>\n",
       "      <th>word_Common</th>\n",
       "      <th>word_Total</th>\n",
       "      <th>word_share</th>\n",
       "      <th>freq_q1+q2</th>\n",
       "      <th>...</th>\n",
       "      <th>cityblock_distance</th>\n",
       "      <th>jaccard_distance</th>\n",
       "      <th>canberra_distance</th>\n",
       "      <th>euclidean_distance</th>\n",
       "      <th>minkowski_distance</th>\n",
       "      <th>braycurtis_distance</th>\n",
       "      <th>skew_q1vec</th>\n",
       "      <th>skew_q2vec</th>\n",
       "      <th>kur_q1vec</th>\n",
       "      <th>kur_q2vec</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>44</td>\n",
       "      <td>47</td>\n",
       "      <td>8</td>\n",
       "      <td>8</td>\n",
       "      <td>4</td>\n",
       "      <td>16</td>\n",
       "      <td>0.250000</td>\n",
       "      <td>2</td>\n",
       "      <td>...</td>\n",
       "      <td>8.690568</td>\n",
       "      <td>1.0</td>\n",
       "      <td>125.797689</td>\n",
       "      <td>0.646990</td>\n",
       "      <td>0.294016</td>\n",
       "      <td>0.322897</td>\n",
       "      <td>0.110333</td>\n",
       "      <td>0.253555</td>\n",
       "      <td>-0.472383</td>\n",
       "      <td>0.007384</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3</td>\n",
       "      <td>17</td>\n",
       "      <td>87</td>\n",
       "      <td>92</td>\n",
       "      <td>17</td>\n",
       "      <td>18</td>\n",
       "      <td>8</td>\n",
       "      <td>33</td>\n",
       "      <td>0.242424</td>\n",
       "      <td>20</td>\n",
       "      <td>...</td>\n",
       "      <td>6.874296</td>\n",
       "      <td>1.0</td>\n",
       "      <td>111.968110</td>\n",
       "      <td>0.508915</td>\n",
       "      <td>0.232597</td>\n",
       "      <td>0.255839</td>\n",
       "      <td>0.100403</td>\n",
       "      <td>-0.047450</td>\n",
       "      <td>-0.011347</td>\n",
       "      <td>-0.036685</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>2 rows × 39 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   freq_qid1  freq_qid2  q1len  q2len  q1_n_words  q2_n_words  word_Common  \\\n",
       "0          1          1     44     47           8           8            4   \n",
       "1          3         17     87     92          17          18            8   \n",
       "\n",
       "   word_Total  word_share  freq_q1+q2  ...  cityblock_distance  \\\n",
       "0          16    0.250000           2  ...            8.690568   \n",
       "1          33    0.242424          20  ...            6.874296   \n",
       "\n",
       "   jaccard_distance  canberra_distance  euclidean_distance  \\\n",
       "0               1.0         125.797689            0.646990   \n",
       "1               1.0         111.968110            0.508915   \n",
       "\n",
       "   minkowski_distance  braycurtis_distance  skew_q1vec  skew_q2vec  kur_q1vec  \\\n",
       "0            0.294016             0.322897    0.110333    0.253555  -0.472383   \n",
       "1            0.232597             0.255839    0.100403   -0.047450  -0.011347   \n",
       "\n",
       "   kur_q2vec  \n",
       "0   0.007384  \n",
       "1  -0.036685  \n",
       "\n",
       "[2 rows x 39 columns]"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "#features= list(data.columns) \n",
    "#print(features)\n",
    "\n",
    "#npmax= np.max(abs(data))\n",
    "#print(npmax)\n",
    "#npmax= np.where(np.isnan(data))\n",
    "#print(npmax)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h2> 5.3 Random train test split( 70:30) </h2>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "        freq_qid1  freq_qid2     q1len     q2len  q1_n_words  q2_n_words  \\\n",
      "0       -0.410741  -0.344466 -0.547312 -0.409319   -0.571549   -0.526573   \n",
      "1        0.030339   2.303063  0.897705  0.916559    1.096633    1.052211   \n",
      "2       -0.410741  -0.344466 -0.480102 -0.114680   -0.386195    0.104941   \n",
      "3       -0.410741  -0.344466 -0.446497 -0.468247   -0.571549   -0.526573   \n",
      "4       -0.410741  -0.344466 -0.580918  0.562991   -0.571549    0.104941   \n",
      "...           ...        ...       ...       ...         ...         ...   \n",
      "196428  -0.190201  -0.344466  3.048430  1.122806    2.394107    0.894333   \n",
      "196429  -0.410741  -0.344466 -0.681733 -0.320928   -0.756902   -0.684451   \n",
      "196430   1.133041   0.317416 -0.043237 -0.291464    0.355219   -0.210816   \n",
      "196431  -0.410741  -0.178996 -1.084994 -0.792351   -1.127609   -0.684451   \n",
      "196432  -0.190201  -0.178996 -0.480102 -0.350391   -0.571549   -0.526573   \n",
      "\n",
      "        word_Common  word_Total  word_share  freq_q1+q2  ...  \\\n",
      "0         -0.187430   -0.632790    0.202507   -0.430560  ...   \n",
      "1          1.099767    1.361168    0.143476    1.534388  ...   \n",
      "2          0.456169   -0.046332    0.480798   -0.430560  ...   \n",
      "3          0.456169   -0.632790    1.176524   -0.430560  ...   \n",
      "4         -0.187430   -0.280915   -0.105077   -0.430560  ...   \n",
      "...             ...         ...         ...         ...  ...   \n",
      "196428    -0.831028    1.361168   -1.273276   -0.321397  ...   \n",
      "196429     0.134370   -0.867374    1.037378   -0.430560  ...   \n",
      "196430     0.456169    0.070960    0.379601    0.770241  ...   \n",
      "196431    -1.152827   -1.101957   -1.096182   -0.321397  ...   \n",
      "196432    -0.187430   -0.632790    0.202507   -0.212233  ...   \n",
      "\n",
      "        cityblock_distance  jaccard_distance  canberra_distance  \\\n",
      "0                -0.110411          0.270944          -0.120435   \n",
      "1                -0.536192          0.270944          -0.412653   \n",
      "2                 0.258544          0.270944           0.310276   \n",
      "3                -0.959362          0.270944          -0.719683   \n",
      "4                 0.376353          0.270944           0.482984   \n",
      "...                    ...               ...                ...   \n",
      "196428            1.035412          0.270944           0.894350   \n",
      "196429           -0.167200          0.270944          -0.063946   \n",
      "196430           -0.389599          0.270944          -0.174143   \n",
      "196431            0.045412          0.270944           0.292318   \n",
      "196432            0.279953          0.270944           0.413238   \n",
      "\n",
      "        euclidean_distance  minkowski_distance  braycurtis_distance  \\\n",
      "0                -0.049398           -0.033946            -0.229084   \n",
      "1                -0.497438           -0.475514            -0.572401   \n",
      "2                 0.276967            0.302532             0.151338   \n",
      "3                -0.958763           -0.954518            -0.935784   \n",
      "4                 0.357814            0.355826             0.298071   \n",
      "...                    ...                 ...                  ...   \n",
      "196428            1.004357            0.979004             0.959993   \n",
      "196429           -0.173842           -0.188724            -0.264218   \n",
      "196430           -0.380405           -0.375080            -0.441413   \n",
      "196431            0.010997            0.001562            -0.020222   \n",
      "196432            0.278893            0.259204             0.215905   \n",
      "\n",
      "        skew_q1vec  skew_q2vec  kur_q1vec  kur_q2vec  \n",
      "0         0.737928    1.800417  -1.580109   0.231636  \n",
      "1         0.664597   -0.430327   0.154148   0.064780  \n",
      "2        -0.687928   -1.658734  -1.270991   0.346946  \n",
      "3        -0.558599   -0.175095   0.397361  -0.036744  \n",
      "4        -0.190415   -0.477304  -0.741107  -0.131531  \n",
      "...            ...         ...        ...        ...  \n",
      "196428   -0.086606    0.815509  -0.128235   1.322224  \n",
      "196429   -0.084786    0.968775  -0.306834  -1.014652  \n",
      "196430   -0.001176    0.659561  -0.836768  -0.343839  \n",
      "196431    0.409469   -0.595490  -0.457181   0.330953  \n",
      "196432   -0.808017    0.162623   1.376918   2.115762  \n",
      "\n",
      "[196433 rows x 39 columns]\n"
     ]
    }
   ],
   "source": [
    "from sklearn.preprocessing import StandardScaler\n",
    "scaler= StandardScaler().fit(data)\n",
    "StandardizeData= scaler.transform(data)\n",
    "StandardizeData= pd.DataFrame(StandardizeData, index= data.index, columns= data.columns)\n",
    "#data = pd.DataFrame(features)#, columns = col_names)\n",
    "\n",
    "print(StandardizeData)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "#npmax= np.where(np.isinf(StandardizeData))\n",
    "#print(npmax)\n",
    "\n",
    "#nan_rows = StandardizeData[StandardizeData.isnull().any(1)]\n",
    "#print (nan_rows)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "#StandardizeData = StandardizeData[pd.notnull(StandardizeData['EPS'])]\n",
    "\n",
    "#StandardizeData = StandardizeData[StandardizeData['EPS'].notna()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train,X_test, y_train, y_test = train_test_split(StandardizeData, y_true, stratify=y_true, test_size=0.3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of data points in train data : (137503, 39)\n",
      "Number of data points in test data : (58930, 39)\n"
     ]
    }
   ],
   "source": [
    "print(\"Number of data points in train data :\",X_train.shape)\n",
    "print(\"Number of data points in test data :\",X_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "---------- Distribution of output variable in train data ----------\n",
      "Class 0:  0.6256808942350349 Class 1:  0.37431910576496513\n",
      "---------- Distribution of output variable in test data ----------\n",
      "Class 0:  0.3743085016120821 Class 1:  0.3743085016120821\n"
     ]
    }
   ],
   "source": [
    "print(\"-\"*10, \"Distribution of output variable in train data\", \"-\"*10)\n",
    "train_distr = Counter(y_train)\n",
    "train_len = len(y_train)\n",
    "print(\"Class 0: \",int(train_distr[0])/train_len,\"Class 1: \", int(train_distr[1])/train_len)\n",
    "print(\"-\"*10, \"Distribution of output variable in test data\", \"-\"*10)\n",
    "test_distr = Counter(y_test)\n",
    "test_len = len(y_test)\n",
    "print(\"Class 0: \",int(test_distr[1])/test_len, \"Class 1: \",int(test_distr[1])/test_len)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h3> 5.4 Plot the confusion_matrix and call this function when we need </h2>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "# This function plots the confusion matrices given y_i, y_i_hat.\n",
    "def plot_confusion_matrix(test_y, predict_y):\n",
    "    C = confusion_matrix(test_y, predict_y)\n",
    "    \n",
    "    A =(((C.T)/(C.sum(axis=1))).T)\n",
    "    \n",
    "    B =(C/C.sum(axis=0))\n",
    "    \n",
    "    plt.figure(figsize=(20,4))\n",
    "    \n",
    "    labels = [1,2]\n",
    "    # representing A in heatmap format\n",
    "    cmap=sns.light_palette(\"blue\")\n",
    "    plt.subplot(1, 3, 1)\n",
    "    sns.heatmap(C, annot=True, cmap=cmap, fmt=\".3f\", xticklabels=labels, yticklabels=labels)\n",
    "    plt.xlabel('Predicted Class')\n",
    "    plt.ylabel('Original Class')\n",
    "    plt.title(\"Confusion matrix\")\n",
    "    \n",
    "    plt.subplot(1, 3, 2)\n",
    "    sns.heatmap(B, annot=True, cmap=cmap, fmt=\".3f\", xticklabels=labels, yticklabels=labels)\n",
    "    plt.xlabel('Predicted Class')\n",
    "    plt.ylabel('Original Class')\n",
    "    plt.title(\"Precision matrix\")\n",
    "    \n",
    "    plt.subplot(1, 3, 3)\n",
    "    # representing B in heatmap format\n",
    "    sns.heatmap(A, annot=True, cmap=cmap, fmt=\".3f\", xticklabels=labels, yticklabels=labels)\n",
    "    plt.xlabel('Predicted Class')\n",
    "    plt.ylabel('Original Class')\n",
    "    plt.title(\"Recall matrix\")\n",
    "    \n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h2> 5.5 Building a random model (Finding worst-case log-loss) </h2>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Log loss on Test Data using Random Model 0.8868882674346896\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x288 with 6 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "predicted_y = np.zeros((test_len,2))\n",
    "for i in range(test_len):\n",
    "    rand_probs = np.random.rand(1,2)\n",
    "    predicted_y[i] = ((rand_probs/sum(sum(rand_probs)))[0])\n",
    "print(\"Log loss on Test Data using Random Model\",log_loss(y_test, predicted_y, eps=1e-15))\n",
    "\n",
    "predicted_y =np.argmax(predicted_y, axis=1)\n",
    "plot_confusion_matrix(y_test, predicted_y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h2> 5.6 Logistic Regression with hyperparameter tuning </h2>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 264,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "For values of alpha =  1e-05 The log loss is: 0.4163610956433371\n",
      "For values of alpha =  0.0001 The log loss is: 0.4089648307889474\n",
      "For values of alpha =  0.001 The log loss is: 0.4089677360420138\n",
      "For values of alpha =  0.01 The log loss is: 0.4248822562082028\n",
      "For values of alpha =  0.1 The log loss is: 0.4646436341579551\n",
      "For values of alpha =  1 The log loss is: 0.5131876989915896\n",
      "For values of alpha =  10 The log loss is: 0.5399326087019145\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "For values of best alpha =  0.0001 The train log loss is: 0.4047327488942188\n",
      "For values of best alpha =  0.0001 The test log loss is: 0.4089648307889474\n",
      "Total number of data points : 58930\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x288 with 6 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "alpha = [10 ** x for x in range(-5, 2)] # hyperparam for SGD classifier.\n",
    "\n",
    "log_error_array=[]\n",
    "for i in alpha:\n",
    "    clf = SGDClassifier(alpha=i, penalty='l2', loss='log', random_state=42)\n",
    "    clf.fit(X_train, y_train)\n",
    "    sig_clf = CalibratedClassifierCV(clf, method=\"sigmoid\")\n",
    "    sig_clf.fit(X_train, y_train)\n",
    "    predict_y = sig_clf.predict_proba(X_test)\n",
    "    log_error_array.append(log_loss(y_test, predict_y, labels=clf.classes_, eps=1e-15))\n",
    "    print('For values of alpha = ', i, \"The log loss is:\",log_loss(y_test, predict_y, labels=clf.classes_, eps=1e-15))\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "ax.plot(alpha, log_error_array,c='g')\n",
    "for i, txt in enumerate(np.round(log_error_array,3)):\n",
    "    ax.annotate((alpha[i],np.round(txt,3)), (alpha[i],log_error_array[i]))\n",
    "plt.grid()\n",
    "plt.title(\"Cross Validation Error for each alpha\")\n",
    "plt.xlabel(\"Alpha i's\")\n",
    "plt.ylabel(\"Error measure\")\n",
    "plt.show()\n",
    "\n",
    "\n",
    "best_alpha = np.argmin(log_error_array)\n",
    "clf = SGDClassifier(alpha=alpha[best_alpha], penalty='l2', loss='log', random_state=42)\n",
    "clf.fit(X_train, y_train)\n",
    "sig_clf = CalibratedClassifierCV(clf, method=\"sigmoid\")\n",
    "sig_clf.fit(X_train, y_train)\n",
    "\n",
    "predict_y = sig_clf.predict_proba(X_train)\n",
    "print('For values of best alpha = ', alpha[best_alpha], \"The train log loss is:\",log_loss(y_train, predict_y, labels=clf.classes_, eps=1e-15))\n",
    "predict_y = sig_clf.predict_proba(X_test)\n",
    "print('For values of best alpha = ', alpha[best_alpha], \"The test log loss is:\",log_loss(y_test, predict_y, labels=clf.classes_, eps=1e-15))\n",
    "predicted_y =np.argmax(predict_y,axis=1)\n",
    "print(\"Total number of data points :\", len(predicted_y))\n",
    "plot_confusion_matrix(y_test, predicted_y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h2> 5.7 Linear SVM with hyperparameter tuning </h2>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 265,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "For values of alpha =  1e-05 The log loss is: 0.42500241257941684\n",
      "For values of alpha =  0.0001 The log loss is: 0.4187505133840204\n",
      "For values of alpha =  0.001 The log loss is: 0.4211436636887012\n",
      "For values of alpha =  0.01 The log loss is: 0.43054913141970663\n",
      "For values of alpha =  0.1 The log loss is: 0.5144980872242297\n",
      "For values of alpha =  1 The log loss is: 0.6612089829555492\n",
      "For values of alpha =  10 The log loss is: 0.661208982955549\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "For values of best alpha =  0.0001 The train log loss is: 0.41546302987316697\n",
      "For values of best alpha =  0.0001 The test log loss is: 0.4187505133840204\n",
      "Total number of data points : 58930\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x288 with 6 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from warnings import simplefilter\n",
    "from sklearn.exceptions import ConvergenceWarning\n",
    "simplefilter(\"ignore\", category=ConvergenceWarning)\n",
    "    \n",
    "\n",
    "alpha = [10 ** x for x in range(-5, 2)] # hyperparam for SGD classifier.\n",
    "\n",
    "log_error_array=[]\n",
    "for i in alpha:\n",
    "    clf = SGDClassifier(alpha=i, penalty='l1', loss='hinge',max_iter=1000, random_state=42)\n",
    "    clf.fit(X_train, y_train)\n",
    "    sig_clf = CalibratedClassifierCV(clf, method=\"sigmoid\")\n",
    "    sig_clf.fit(X_train, y_train)\n",
    "    predict_y = sig_clf.predict_proba(X_test)\n",
    "    log_error_array.append(log_loss(y_test, predict_y, labels=clf.classes_, eps=1e-15))\n",
    "    print('For values of alpha = ', i, \"The log loss is:\",log_loss(y_test, predict_y, labels=clf.classes_, eps=1e-15))\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "ax.plot(alpha, log_error_array,c='g')\n",
    "for i, txt in enumerate(np.round(log_error_array,3)):\n",
    "    ax.annotate((alpha[i],np.round(txt,3)), (alpha[i],log_error_array[i]))\n",
    "plt.grid()\n",
    "plt.title(\"Cross Validation Error for each alpha\")\n",
    "plt.xlabel(\"Alpha i's\")\n",
    "plt.ylabel(\"Error measure\")\n",
    "plt.show()\n",
    "\n",
    "\n",
    "best_alpha = np.argmin(log_error_array)\n",
    "clf = SGDClassifier(alpha=alpha[best_alpha], penalty='l1', loss='hinge', random_state=42)\n",
    "clf.fit(X_train, y_train)\n",
    "sig_clf = CalibratedClassifierCV(clf, method=\"sigmoid\")\n",
    "sig_clf.fit(X_train, y_train)\n",
    "\n",
    "predict_y = sig_clf.predict_proba(X_train)\n",
    "print('For values of best alpha = ', alpha[best_alpha], \"The train log loss is:\",log_loss(y_train, predict_y, labels=clf.classes_, eps=1e-15))\n",
    "predict_y = sig_clf.predict_proba(X_test)\n",
    "print('For values of best alpha = ', alpha[best_alpha], \"The test log loss is:\",log_loss(y_test, predict_y, labels=clf.classes_, eps=1e-15))\n",
    "predicted_y =np.argmax(predict_y,axis=1)\n",
    "print(\"Total number of data points :\", len(predicted_y))\n",
    "plot_confusion_matrix(y_test, predicted_y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h2> 5.8 Applying XGBoost with hyperparameter tuning using RandomsearchCV </h2>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "## For Hyper Parameter Optimization\n",
    "\n",
    "param_grid = {\n",
    "        'max_depth': [6, 10, 15, 20],\n",
    "        'learning_rate': [0.001, 0.01, 0.1, 0.2, 0.3, 0.5],\n",
    "        'subsample': [0.5, 0.6, 0.7, 0.8, 0.9, 1.0],\n",
    "        'colsample_bytree': [0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0],\n",
    "        'min_child_weight': [0.5, 1.0, 3.0, 5.0, 7.0, 10.0],\n",
    "        'gamma': [0, 0.25, 0.5, 1.0],\n",
    "        'n_estimators': [50,100,150,200,300,400,500]}\n",
    "\n",
    "def timer(start_time=None):\n",
    "    if not start_time:\n",
    "        start_time = datetime.now()\n",
    "        return start_time\n",
    "    elif start_time:\n",
    "        thour, temp_sec = divmod((datetime.now() - start_time).total_seconds(), 3600)\n",
    "        tmin, tsec = divmod(temp_sec, 60)\n",
    "        print('\\n Time taken: %i hours %i minutes and %s seconds.' % (thour, tmin, round(tsec, 2)))\n",
    "\n",
    "        \n",
    "clf = xgb.XGBClassifier(nthread=-1)\n",
    "\n",
    "kfold = StratifiedKFold(n_splits=5, shuffle=True)\n",
    "random_search=RandomizedSearchCV(clf,param_distributions=param_grid,n_iter=5,scoring='neg_log_loss',n_jobs=-1,cv=kfold,verbose=3,random_state=42)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 5 candidates, totalling 25 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 12 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done  11 out of  25 | elapsed: 11.9min remaining: 15.2min\n",
      "[Parallel(n_jobs=-1)]: Done  20 out of  25 | elapsed: 14.5min remaining:  3.6min\n",
      "[Parallel(n_jobs=-1)]: Done  25 out of  25 | elapsed: 21.2min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      " Time taken: 0 hours 22 minutes and 36.8 seconds.\n"
     ]
    }
   ],
   "source": [
    "from datetime import datetime\n",
    "# Here we go\n",
    "start_time = timer(None) # timing starts from this point for \"start_time\" variable\n",
    "\n",
    "random_search.fit(X_train, y_train)\n",
    "\n",
    "timer(start_time) # timing ends here for \"start_time\" variable"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best Score is: -0.336275 from using these combination of param_grid: {'subsample': 1.0, 'n_estimators': 400, 'min_child_weight': 7.0, 'max_depth': 15, 'learning_rate': 0.01, 'gamma': 0, 'colsample_bytree': 1.0}\n",
      "\n",
      "mean : -0.543048, std : (0.001112) with param_grid : {'subsample': 0.7, 'n_estimators': 400, 'min_child_weight': 10.0, 'max_depth': 20, 'learning_rate': 0.001, 'gamma': 0, 'colsample_bytree': 0.9}\n",
      "mean : -0.336275, std : (0.003963) with param_grid : {'subsample': 1.0, 'n_estimators': 400, 'min_child_weight': 7.0, 'max_depth': 15, 'learning_rate': 0.01, 'gamma': 0, 'colsample_bytree': 1.0}\n",
      "mean : -0.386254, std : (0.007723) with param_grid : {'subsample': 0.9, 'n_estimators': 100, 'min_child_weight': 5.0, 'max_depth': 20, 'learning_rate': 0.3, 'gamma': 0.25, 'colsample_bytree': 0.9}\n",
      "mean : -0.517357, std : (0.001350) with param_grid : {'subsample': 0.7, 'n_estimators': 500, 'min_child_weight': 3.0, 'max_depth': 20, 'learning_rate': 0.001, 'gamma': 0.25, 'colsample_bytree': 0.8}\n",
      "mean : -0.370329, std : (0.007633) with param_grid : {'subsample': 1.0, 'n_estimators': 100, 'min_child_weight': 7.0, 'max_depth': 20, 'learning_rate': 0.3, 'gamma': 1.0, 'colsample_bytree': 0.8}\n"
     ]
    }
   ],
   "source": [
    "# Summarize results\n",
    "print(\"Best Score is: %f from using these combination of param_grid: %s\"%(random_search.best_score_,random_search.best_params_))\n",
    "print()\n",
    "means = random_search.cv_results_['mean_test_score']\n",
    "stds = random_search.cv_results_['std_test_score']\n",
    "params = random_search.cv_results_['params']\n",
    "for mean, stdev, param in zip(means, stds, params):\n",
    "    print(\"mean : %f, std : (%f) with param_grid : %r\" % (mean, stdev, param))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "XGBClassifier(base_score=0.5, booster='gbtree', colsample_bylevel=1,\n",
       "              colsample_bynode=1, colsample_bytree=1.0, gamma=0, gpu_id=-1,\n",
       "              importance_type='gain', interaction_constraints='',\n",
       "              learning_rate=0.01, max_delta_step=0, max_depth=15,\n",
       "              min_child_weight=7.0, missing=nan, monotone_constraints='()',\n",
       "              n_estimators=400, n_jobs=-1, nthread=-1, num_parallel_tree=1,\n",
       "              random_state=0, reg_alpha=0, reg_lambda=1, scale_pos_weight=1,\n",
       "              subsample=1.0, tree_method='exact', validate_parameters=1,\n",
       "              verbosity=None)"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "random_search.best_estimator_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'subsample': 1.0,\n",
       " 'n_estimators': 400,\n",
       " 'min_child_weight': 7.0,\n",
       " 'max_depth': 15,\n",
       " 'learning_rate': 0.01,\n",
       " 'gamma': 0,\n",
       " 'colsample_bytree': 1.0}"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "random_search.best_params_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<html>\n",
    "<body><b>\n",
    "<p style=\"color: #0099cc\">SO, After Applying XGBoost with hyperparameter tuning using RandomsearchCV I got the best parameter for best score. Now, I am going to apply this best parameter to XGBClassifier to getting best logg loss score.</p>\n",
    "</b></body>\n",
    "</html>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The train log loss is: 0.2465866826928993\n",
      "The test log loss is: 0.3351238484184\n",
      "Total number of data points : 58930\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x288 with 6 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "clf=xgb.XGBClassifier(n_jobs=-1,random_state=42,\n",
    "                      max_depth=15,\n",
    "                      n_estimators=400,\n",
    "                      subsample= 1.0, \n",
    "                      min_child_weight= 7.0,\n",
    "                      learning_rate= 0.01,\n",
    "                      gamma= 0,\n",
    "                      colsample_bytree= 1.0)\n",
    "clf.fit(X_train,y_train)\n",
    "\n",
    "predict_y = clf.predict_proba(X_train)\n",
    "print(\"The train log loss is:\",log_loss(y_train, predict_y, labels=clf.classes_, eps=1e-15))\n",
    "predict_y = clf.predict_proba(X_test)\n",
    "print(\"The test log loss is:\",log_loss(y_test, predict_y, labels=clf.classes_, eps=1e-15))\n",
    "\n",
    "predicted_y =np.argmax(predict_y,axis=1)\n",
    "print(\"Total number of data points :\", len(predicted_y))\n",
    "plot_confusion_matrix(y_test, predicted_y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}