question1_1.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This is part of the Question 1. Here you can find the AdaBoostRegressor implementation with output "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.ensemble import RandomForestRegressor\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.model_selection import train_test_split\n",
    "from catboost import Pool, CatBoostRegressor, cv"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The shape of our features is: (120915, 10)\n"
     ]
    }
   ],
   "source": [
    "df=pd.read_csv(\"dataprocessed.csv\")\n",
    "print('The shape of our features is:', df.shape)"
   ]
  },
  {
   "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></th>\n",
       "      <th>title</th>\n",
       "      <th>variety</th>\n",
       "      <th>designation</th>\n",
       "      <th>country</th>\n",
       "      <th>province</th>\n",
       "      <th>description</th>\n",
       "      <th>winery</th>\n",
       "      <th>taster_name</th>\n",
       "      <th>points</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>quinta dos avidagos 2011 avidagos red (douro)</td>\n",
       "      <td>portuguese red</td>\n",
       "      <td>avidagos</td>\n",
       "      <td>portugal</td>\n",
       "      <td>douro</td>\n",
       "      <td>this is ripe and fruity, a wine that is smooth...</td>\n",
       "      <td>quinta dos avidagos</td>\n",
       "      <td>roger voss</td>\n",
       "      <td>87</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>rainstorm 2013 pinot gris (willamette valley)</td>\n",
       "      <td>pinot gris</td>\n",
       "      <td>unknown</td>\n",
       "      <td>us</td>\n",
       "      <td>oregon</td>\n",
       "      <td>tart and snappy, the flavors of lime flesh and...</td>\n",
       "      <td>rainstorm</td>\n",
       "      <td>paul gregutt</td>\n",
       "      <td>87</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>st. julian 2013 reserve late harvest riesling ...</td>\n",
       "      <td>riesling</td>\n",
       "      <td>reserve late harvest</td>\n",
       "      <td>us</td>\n",
       "      <td>michigan</td>\n",
       "      <td>pineapple rind, lemon pith and orange blossom ...</td>\n",
       "      <td>st. julian</td>\n",
       "      <td>alexander peartree</td>\n",
       "      <td>87</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>sweet cheeks 2012 vintner's reserve wild child...</td>\n",
       "      <td>pinot noir</td>\n",
       "      <td>vintner's reserve wild child block</td>\n",
       "      <td>us</td>\n",
       "      <td>oregon</td>\n",
       "      <td>much like the regular bottling from 2012, this...</td>\n",
       "      <td>sweet cheeks</td>\n",
       "      <td>paul gregutt</td>\n",
       "      <td>87</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>tandem 2011 ars in vitro tempranillo-merlot (n...</td>\n",
       "      <td>tempranillo-merlot</td>\n",
       "      <td>ars in vitro</td>\n",
       "      <td>spain</td>\n",
       "      <td>northern spain</td>\n",
       "      <td>blackberry and raspberry aromas show a typical...</td>\n",
       "      <td>tandem</td>\n",
       "      <td>michael schachner</td>\n",
       "      <td>87</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                               title             variety  \\\n",
       "0      quinta dos avidagos 2011 avidagos red (douro)      portuguese red   \n",
       "1      rainstorm 2013 pinot gris (willamette valley)          pinot gris   \n",
       "2  st. julian 2013 reserve late harvest riesling ...            riesling   \n",
       "3  sweet cheeks 2012 vintner's reserve wild child...          pinot noir   \n",
       "4  tandem 2011 ars in vitro tempranillo-merlot (n...  tempranillo-merlot   \n",
       "\n",
       "                          designation   country        province  \\\n",
       "0                            avidagos  portugal           douro   \n",
       "1                             unknown        us          oregon   \n",
       "2                reserve late harvest        us        michigan   \n",
       "3  vintner's reserve wild child block        us          oregon   \n",
       "4                        ars in vitro     spain  northern spain   \n",
       "\n",
       "                                         description               winery  \\\n",
       "0  this is ripe and fruity, a wine that is smooth...  quinta dos avidagos   \n",
       "1  tart and snappy, the flavors of lime flesh and...            rainstorm   \n",
       "2  pineapple rind, lemon pith and orange blossom ...           st. julian   \n",
       "3  much like the regular bottling from 2012, this...         sweet cheeks   \n",
       "4  blackberry and raspberry aromas show a typical...               tandem   \n",
       "\n",
       "          taster_name  points  \n",
       "0          roger voss      87  \n",
       "1        paul gregutt      87  \n",
       "2  alexander peartree      87  \n",
       "3        paul gregutt      87  \n",
       "4   michael schachner      87  "
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "labels=np.array(df['price'])\n",
    "df=df.drop('price',axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Index(['variety', 'designation', 'country', 'province', 'winery',\n",
      "       'taster_name', 'points'],\n",
      "      dtype='object')\n"
     ]
    }
   ],
   "source": [
    "X=df[['variety', 'designation', 'country', 'province',\n",
    "       'winery', 'taster_name', 'points']]\n",
    "\n",
    "X=X.fillna('unknown')\n",
    "print(X.columns)\n",
    "categorical_features_indices =[0,1,2,3,4,5]\n",
    "y=labels\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, \n",
    "                                                    random_state=42)\n",
    "\n",
    "X_train, X_valid, y_train, y_valid = train_test_split(X_train, y_train, test_size=0.2, \n",
    "                                                    random_state=52)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[0, 1, 2, 3, 4, 5]"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "categorical_features_indices"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [],
   "source": [
    "def perform_model(X_train, y_train,X_valid, y_valid,X_test, y_test):\n",
    "    model = CatBoostRegressor(\n",
    "        random_seed = 400,\n",
    "        loss_function = 'RMSE',\n",
    "        iterations=100,\n",
    "    )\n",
    "    \n",
    "    model.fit(\n",
    "        X_train, y_train,\n",
    "        cat_features = categorical_features_indices,\n",
    "        eval_set=(X_valid, y_valid),\n",
    "        verbose=False\n",
    "    )\n",
    "    \n",
    "    print(\"RMSE on training data: \"+ model.score(X_train, y_train).astype(str))\n",
    "    print(\"RMSE on test data: \"+ model.score(X_test, y_test).astype(str))\n",
    "    \n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "RMSE on training data: 0.5591728787094588\n",
      "RMSE on test data: 0.3514526190642865\n"
     ]
    }
   ],
   "source": [
    "model=perform_model(X_train, y_train,X_valid, y_valid,X_test, y_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can see model achieved RMSE(root mean square error) error 0.35 for test set "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [],
   "source": [
    "feature_score = pd.DataFrame(list(zip(X.dtypes.index, model.get_feature_importance(Pool(X, label=y, cat_features=categorical_features_indices)))),\n",
    "                columns=['Feature','Score'])\n",
    "\n",
    "feature_score = feature_score.sort_values(by='Score', ascending=False, inplace=False, kind='quicksort', na_position='last')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {},
   "outputs": [
    {
     "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.rcParams[\"figure.figsize\"] = (12,7)\n",
    "ax = feature_score.plot('Feature', 'Score', kind='bar', color='c')\n",
    "ax.set_title(\"Catboost Feature Importance Ranking\", fontsize = 14)\n",
    "ax.set_xlabel('')\n",
    "\n",
    "rects = ax.patches\n",
    "\n",
    "labels = feature_score['Score'].round(2)\n",
    "\n",
    "for rect, label in zip(rects, labels):\n",
    "    height = rect.get_height()\n",
    "    ax.text(rect.get_x() + rect.get_width()/2, height + 0.35, label, ha='center', va='bottom')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The plot shows where are the important features for our problem. As we expected points are very important and country is not very important as feature."
   ]
  },
  {
   "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": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}