{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Car Price"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-17T23:20:54.071563Z",
     "start_time": "2021-09-17T23:20:41.130582Z"
    },
    "id": "cS-op1_XQzR4"
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "id": "zT5EsjTLU71z"
   },
   "outputs": [],
   "source": [
    "df_test= pd.read_csv(\"test_car_details.csv\")\n",
    "df_train = pd.read_csv(\"train_car_details.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "YMjgguNTjtxK",
    "outputId": "00a9fc66-1681-4294-ae9c-22f277486ef7"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Index(['Id', 'name', 'year', 'selling_price', 'km_driven', 'fuel',\n",
      "       'seller_type', 'transmission', 'owner', 'mileage', 'engine',\n",
      "       'max_power', 'torque', 'seats'],\n",
      "      dtype='object')\n",
      "Index(['Id', 'name', 'year', 'km_driven', 'fuel', 'seller_type',\n",
      "       'transmission', 'owner', 'mileage', 'engine', 'max_power', 'torque',\n",
      "       'seats'],\n",
      "      dtype='object')\n"
     ]
    }
   ],
   "source": [
    "print(df_train.columns)\n",
    "print(df_test.columns)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "nIMoUKctL0xx"
   },
   "source": [
    "## Objetivo"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Rvi3gtTaSqdU"
   },
   "source": [
    "Um dos problemas que ocorrem na OLX da Índia, pela baixa volumetria de dados, é a empresa não conseguir estimar um preço de venda para o carro do seu cliente baseado em algumas características do veículo. O objetivo é estimar tal valor a partir de dados do concorrente (CarDekho)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "nKnbPT9VUKmw"
   },
   "source": [
    "## Análise qualitativa e quantitativa dos dados"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 289
    },
    "id": "vSgDhqnuUaxY",
    "outputId": "147325c1-0008-4356-c041-f79d79c9acd2"
   },
   "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>name</th>\n",
       "      <th>year</th>\n",
       "      <th>selling_price</th>\n",
       "      <th>km_driven</th>\n",
       "      <th>fuel</th>\n",
       "      <th>seller_type</th>\n",
       "      <th>transmission</th>\n",
       "      <th>owner</th>\n",
       "      <th>mileage</th>\n",
       "      <th>engine</th>\n",
       "      <th>max_power</th>\n",
       "      <th>torque</th>\n",
       "      <th>seats</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>Hyundai Santro GLS I - Euro I</td>\n",
       "      <td>1999</td>\n",
       "      <td>80000</td>\n",
       "      <td>110000</td>\n",
       "      <td>Petrol</td>\n",
       "      <td>Individual</td>\n",
       "      <td>Manual</td>\n",
       "      <td>Second Owner</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>1</th>\n",
       "      <td>2</td>\n",
       "      <td>Maruti Ertiga VDI</td>\n",
       "      <td>2012</td>\n",
       "      <td>459999</td>\n",
       "      <td>87000</td>\n",
       "      <td>Diesel</td>\n",
       "      <td>Individual</td>\n",
       "      <td>Manual</td>\n",
       "      <td>First Owner</td>\n",
       "      <td>20.77 kmpl</td>\n",
       "      <td>1248 CC</td>\n",
       "      <td>88.76 bhp</td>\n",
       "      <td>200Nm@ 1750rpm</td>\n",
       "      <td>7.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>BMW 3 Series 320d Luxury Line</td>\n",
       "      <td>2010</td>\n",
       "      <td>1100000</td>\n",
       "      <td>102000</td>\n",
       "      <td>Diesel</td>\n",
       "      <td>Dealer</td>\n",
       "      <td>Automatic</td>\n",
       "      <td>First Owner</td>\n",
       "      <td>19.62 kmpl</td>\n",
       "      <td>1995 CC</td>\n",
       "      <td>187.74 bhp</td>\n",
       "      <td>400Nm@ 1750-2500rpm</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>Tata New Safari DICOR 2.2 EX 4x2</td>\n",
       "      <td>2009</td>\n",
       "      <td>229999</td>\n",
       "      <td>212000</td>\n",
       "      <td>Diesel</td>\n",
       "      <td>Individual</td>\n",
       "      <td>Manual</td>\n",
       "      <td>Third Owner</td>\n",
       "      <td>11.57 kmpl</td>\n",
       "      <td>2179 CC</td>\n",
       "      <td>138.1 bhp</td>\n",
       "      <td>320Nm@ 1700-2700rpm</td>\n",
       "      <td>7.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>Toyota Fortuner 3.0 Diesel</td>\n",
       "      <td>2010</td>\n",
       "      <td>800000</td>\n",
       "      <td>125000</td>\n",
       "      <td>Diesel</td>\n",
       "      <td>Individual</td>\n",
       "      <td>Manual</td>\n",
       "      <td>Second Owner</td>\n",
       "      <td>11.5 kmpl</td>\n",
       "      <td>2982 CC</td>\n",
       "      <td>171 bhp</td>\n",
       "      <td>343Nm@ 1400-3400rpm</td>\n",
       "      <td>7.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Id                              name  year  selling_price  km_driven  \\\n",
       "0   1     Hyundai Santro GLS I - Euro I  1999          80000     110000   \n",
       "1   2                 Maruti Ertiga VDI  2012         459999      87000   \n",
       "2   3     BMW 3 Series 320d Luxury Line  2010        1100000     102000   \n",
       "3   4  Tata New Safari DICOR 2.2 EX 4x2  2009         229999     212000   \n",
       "4   5        Toyota Fortuner 3.0 Diesel  2010         800000     125000   \n",
       "\n",
       "     fuel seller_type transmission         owner     mileage   engine  \\\n",
       "0  Petrol  Individual       Manual  Second Owner         NaN      NaN   \n",
       "1  Diesel  Individual       Manual   First Owner  20.77 kmpl  1248 CC   \n",
       "2  Diesel      Dealer    Automatic   First Owner  19.62 kmpl  1995 CC   \n",
       "3  Diesel  Individual       Manual   Third Owner  11.57 kmpl  2179 CC   \n",
       "4  Diesel  Individual       Manual  Second Owner   11.5 kmpl  2982 CC   \n",
       "\n",
       "    max_power               torque  seats  \n",
       "0         NaN                  NaN    NaN  \n",
       "1   88.76 bhp       200Nm@ 1750rpm    7.0  \n",
       "2  187.74 bhp  400Nm@ 1750-2500rpm    5.0  \n",
       "3   138.1 bhp  320Nm@ 1700-2700rpm    7.0  \n",
       "4     171 bhp  343Nm@ 1400-3400rpm    7.0  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Verificando as variáveis\n",
    "df_train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "id": "DXkxkyyDa2aj"
   },
   "outputs": [],
   "source": [
    "df_train = df_train.iloc[:,1:] #retirando a coluna de id\n",
    "\n",
    "Id = df_test.Id\n",
    "df_test = df_test.iloc[:,1:] #retirando a coluna de id"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "O8lv_YnOO_6P"
   },
   "source": [
    "### Analise de dados nulos"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "zP8597KtW7pX",
    "outputId": "43539f2c-5b17-4da0-885d-678841f07fa3"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "name               0\n",
       "year               0\n",
       "selling_price      0\n",
       "km_driven          0\n",
       "fuel               0\n",
       "seller_type        0\n",
       "transmission       0\n",
       "owner              0\n",
       "mileage          157\n",
       "engine           157\n",
       "max_power        151\n",
       "torque           158\n",
       "seats            157\n",
       "dtype: int64"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Verificado a quantidade de NaN por atributo\n",
    "df_train.isna().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "EtZj5WU1PME8",
    "outputId": "3f68afd7-1bf8-477e-cc13-e6a615c3b00d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "name             0.000000\n",
      "year             0.000000\n",
      "selling_price    0.000000\n",
      "km_driven        0.000000\n",
      "fuel             0.000000\n",
      "seller_type      0.000000\n",
      "transmission     0.000000\n",
      "owner            0.000000\n",
      "mileage          2.759712\n",
      "engine           2.759712\n",
      "max_power        2.654245\n",
      "torque           2.777290\n",
      "seats            2.759712\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "#Porcentagem de nan por atributo\n",
    "print(100*df_train.isna().sum()/len(df_train))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "DghC0QGiq_np",
    "outputId": "15cd8e5a-73a1-45d9-aeab-a17a36e4ea81"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "name            0.0\n",
      "year            0.0\n",
      "km_driven       0.0\n",
      "fuel            0.0\n",
      "seller_type     0.0\n",
      "transmission    0.0\n",
      "owner           0.0\n",
      "mileage         0.0\n",
      "engine          0.0\n",
      "max_power       0.0\n",
      "torque          0.0\n",
      "seats           0.0\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "#Porcentagem de nan por atributo no test\n",
    "print(100*df_test.isna().sum()/len(df_test))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "cPeACfqLXdCL"
   },
   "source": [
    "Pelo fatos dos NaN's estar majoritariamente presente nas mesmas linhas e por representar um baixo volume em relação ao total (menos de 3%), tais linhas serão retiradas."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "SLNlNI1vYz3g",
    "outputId": "e32eb487-fbe8-4c06-df6f-2c5eb59c239a"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Quantidade de linhas totais:  5689\n",
      "Quantidade de linhas após retirada dos NaNs:  5531\n"
     ]
    }
   ],
   "source": [
    "print(f'Quantidade de linhas totais: ', df_train.shape[0])\n",
    "# Remove as linhas com NaN\n",
    "df_train = df_train.dropna(axis=0)   \n",
    "print(f'Quantidade de linhas após retirada dos NaNs: ', df_train.shape[0])\n",
    "#Aproximadamente 3% de linhas eliminadas"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "2MhTDUWnQlvV"
   },
   "source": [
    "### Categoria das variaveis"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "jLjrd98qlWMR"
   },
   "source": [
    "Os dados são compostos pelas variáveis:\n",
    "\n",
    "\n",
    "* Variaveis quantitativas discreta:<br>\n",
    " * Ano de fabricacao do carro (year)\n",
    " * Qtd de Km dirigidos (km_driven)\n",
    " * Potência máxima do motor (max_power)\n",
    " * Qtd de acentos (seats)\n",
    "* Variaveis quantitativas continuas:\n",
    " * Quilometragem por litro (mileage)\n",
    " * Potencia do motor (engine)\n",
    " * Preço de venda (selling_price) **Valor a ser predito**\n",
    "* Variaveis qualitativas nominais:\n",
    " * nome do carro (name)\n",
    " * tipo de combustivel utilizado (fuel)\n",
    " * tipo de vendendor (seller_type)\n",
    " * transmissao (transmission)\n",
    " * Torque: responsável pela capacidade do motor produzir força motriz, ou seja, o movimento giratório\n",
    "* Variaveis qualitativas ordinais:\n",
    " * Quantos donos ja possuiram o carro (owner)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "qX0b_8ToVmFq"
   },
   "source": [
    "Nota-se que 4 variáveis são numéricas, mas é necessário uma tratativa para retirar as strings que representam a unidade de medida. Ao todo, considerando as variáveis que precisam ser tratadas, há 7 variáveis numéricas e 6 variáveis categóricas qualitativas."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "1E4O0qOfncSS"
   },
   "source": [
    "### Retirando a palavra owner da coluna owner, retirando a unidade de medida de mileage, engine e max power e retirando a segunda unidade de medida utilizada no torque (rpm) e deixando apenas a unidade Nm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "id": "3PJ68h7Jp0SC"
   },
   "outputs": [],
   "source": [
    "df1 = df_train.copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "id": "Lyn3uW_sxna0"
   },
   "outputs": [],
   "source": [
    "colunas = ['owner', 'mileage', 'engine',\n",
    "       'max_power', 'torque']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "id": "HKyNC-IdyKVT"
   },
   "outputs": [],
   "source": [
    "#base de treino\n",
    "for i in colunas:\n",
    "  df1[i] = df1[i].str.split(' ').str[0]\n",
    "\n",
    "#Na base de test\n",
    "for i in colunas:\n",
    "  df_test[i] = df_test[i].str.split(' ').str[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "HEWJoQeioxpp"
   },
   "source": [
    "### Retirando a unidade de medida do torque"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "id": "MjhqBrFRcqIJ"
   },
   "outputs": [],
   "source": [
    "#Retirando a unidade de medida do torque \n",
    "df1['torque'] = df1['torque'].str.replace('Nm@', '', regex=True).replace('nm@', '', regex=True).replace('@', '', regex=True).replace('Nm', '', regex=True).replace('NM', '', regex=True).replace('kgm', '', regex=True)\n",
    "\n",
    "#no test\n",
    "df_test['torque'] = df_test['torque'].str.replace('Nm@', '', regex=True).replace('nm@', '', regex=True).replace('@', '', regex=True).replace('Nm', '', regex=True).replace('NM', '', regex=True).replace('kgm', '', regex=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 80
    },
    "id": "fLkK6JyRwC4c",
    "outputId": "11aeb842-76c6-46cc-d22d-d8eaacfed914"
   },
   "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>name</th>\n",
       "      <th>year</th>\n",
       "      <th>selling_price</th>\n",
       "      <th>km_driven</th>\n",
       "      <th>fuel</th>\n",
       "      <th>seller_type</th>\n",
       "      <th>transmission</th>\n",
       "      <th>owner</th>\n",
       "      <th>mileage</th>\n",
       "      <th>engine</th>\n",
       "      <th>max_power</th>\n",
       "      <th>torque</th>\n",
       "      <th>seats</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1954</th>\n",
       "      <td>Honda Jazz Select Edition Active</td>\n",
       "      <td>2011</td>\n",
       "      <td>350000</td>\n",
       "      <td>80000</td>\n",
       "      <td>Petrol</td>\n",
       "      <td>Individual</td>\n",
       "      <td>Manual</td>\n",
       "      <td>Second</td>\n",
       "      <td>16.0</td>\n",
       "      <td>1198</td>\n",
       "      <td>90</td>\n",
       "      <td>110(11.2)</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                  name  year  selling_price  km_driven  \\\n",
       "1954  Honda Jazz Select Edition Active  2011         350000      80000   \n",
       "\n",
       "        fuel seller_type transmission   owner mileage engine max_power  \\\n",
       "1954  Petrol  Individual       Manual  Second    16.0   1198        90   \n",
       "\n",
       "         torque  seats  \n",
       "1954  110(11.2)    5.0  "
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df1[df1['torque'] == '110(11.2)']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "K4aVGdITrz6A"
   },
   "source": [
    "Nao temos esse problema na base de teste"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "D-3Zv5-ZpZoh"
   },
   "source": [
    "#### Tratando a linha com torque (11.2) no treino"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "id": "xte0Egxcyveu"
   },
   "outputs": [],
   "source": [
    "#eliminando a linha com toque = 110(11.2)\n",
    "df1.drop(df1.loc[df1['torque'] == '110(11.2)'].index, inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "rSGU10uWswh3"
   },
   "source": [
    "#### Tratando linha 380(38.7) no torque (base teste)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 80
    },
    "id": "xPJSLw2asuuu",
    "outputId": "d41e780d-4dab-4faa-9d40-e143cdd31d90"
   },
   "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>name</th>\n",
       "      <th>year</th>\n",
       "      <th>km_driven</th>\n",
       "      <th>fuel</th>\n",
       "      <th>seller_type</th>\n",
       "      <th>transmission</th>\n",
       "      <th>owner</th>\n",
       "      <th>mileage</th>\n",
       "      <th>engine</th>\n",
       "      <th>max_power</th>\n",
       "      <th>torque</th>\n",
       "      <th>seats</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>885</th>\n",
       "      <td>Ford Endeavour Hurricane Limited Edition</td>\n",
       "      <td>2013</td>\n",
       "      <td>110000</td>\n",
       "      <td>Diesel</td>\n",
       "      <td>Individual</td>\n",
       "      <td>Automatic</td>\n",
       "      <td>Third</td>\n",
       "      <td>12.8</td>\n",
       "      <td>2953</td>\n",
       "      <td>156</td>\n",
       "      <td>380(38.7)</td>\n",
       "      <td>7.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                         name  year  km_driven    fuel  \\\n",
       "885  Ford Endeavour Hurricane Limited Edition  2013     110000  Diesel   \n",
       "\n",
       "    seller_type transmission  owner mileage engine max_power     torque  seats  \n",
       "885  Individual    Automatic  Third    12.8   2953       156  380(38.7)    7.0  "
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_test[df_test['torque'] == '380(38.7)']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "3fPNDEkcyB1U"
   },
   "source": [
    "Nao podemos eliminar linhas da base de teste\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 80
    },
    "id": "7HKaZDOutACC",
    "outputId": "1e9ccc6c-c15d-44f8-b01e-2e515d9e7831"
   },
   "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>name</th>\n",
       "      <th>year</th>\n",
       "      <th>km_driven</th>\n",
       "      <th>fuel</th>\n",
       "      <th>seller_type</th>\n",
       "      <th>transmission</th>\n",
       "      <th>owner</th>\n",
       "      <th>mileage</th>\n",
       "      <th>engine</th>\n",
       "      <th>max_power</th>\n",
       "      <th>torque</th>\n",
       "      <th>seats</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>885</th>\n",
       "      <td>Ford Endeavour Hurricane Limited Edition</td>\n",
       "      <td>2013</td>\n",
       "      <td>110000</td>\n",
       "      <td>Diesel</td>\n",
       "      <td>Individual</td>\n",
       "      <td>Automatic</td>\n",
       "      <td>Third</td>\n",
       "      <td>12.8</td>\n",
       "      <td>2953</td>\n",
       "      <td>156</td>\n",
       "      <td>380(38.7)</td>\n",
       "      <td>7.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                         name  year  km_driven    fuel  \\\n",
       "885  Ford Endeavour Hurricane Limited Edition  2013     110000  Diesel   \n",
       "\n",
       "    seller_type transmission  owner mileage engine max_power     torque  seats  \n",
       "885  Individual    Automatic  Third    12.8   2953       156  380(38.7)    7.0  "
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_test[df_test['torque'] == '380(38.7)']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "id": "GMBQG03SzPAW"
   },
   "outputs": [],
   "source": [
    "df_test.torque.replace('380(38.7)', '380', inplace= True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 49
    },
    "id": "MZEIkXmWs9No",
    "outputId": "28560560-4a0a-4c39-9ca4-fc3631b8848f"
   },
   "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>name</th>\n",
       "      <th>year</th>\n",
       "      <th>km_driven</th>\n",
       "      <th>fuel</th>\n",
       "      <th>seller_type</th>\n",
       "      <th>transmission</th>\n",
       "      <th>owner</th>\n",
       "      <th>mileage</th>\n",
       "      <th>engine</th>\n",
       "      <th>max_power</th>\n",
       "      <th>torque</th>\n",
       "      <th>seats</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "Empty DataFrame\n",
       "Columns: [name, year, km_driven, fuel, seller_type, transmission, owner, mileage, engine, max_power, torque, seats]\n",
       "Index: []"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_test[df_test['torque'] == '380(38.7)']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "dgAw8_SktMk4"
   },
   "source": [
    "#### Analisando nome dos carros"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "id": "CgRJNSmz-tGZ"
   },
   "outputs": [],
   "source": [
    "#Colocando a marca e modelo do carro em um dicionario juntamente com sua frequência no conjunto \n",
    "import collections\n",
    "agrupamento = df1['name']\n",
    "counter=collections.Counter(agrupamento)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "ytRnpNuj_fhA",
    "outputId": "dd1a59f2-3e5f-47d3-e3fb-14dd06f06a61"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Quantidade de modelos de carros distintos:  1706\n",
      "A maior quantidade de um único modelo de carro:  92\n"
     ]
    }
   ],
   "source": [
    "#Colocando a frequencia em uma lista para poder contar a qtd de itens diferentes\n",
    "contador = []\n",
    "for i in sorted(counter, key = counter.get, reverse = True):\n",
    "    contador.append(counter[i])\n",
    "print(\"Quantidade de modelos de carros distintos: \",len(contador))\n",
    "print(\"A maior quantidade de um único modelo de carro: \",max(contador))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "0f3mIjEDdie7"
   },
   "source": [
    "Como o nome dos carros é muito variado, não é interessante estar presente no modelo, porém, a informação da marca do carro pode ser importante, assim como outras informações oriundas da própria base."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "mRrCRfsLd18B"
   },
   "source": [
    "## Feature engineering"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "sbV_L_k5qi2w"
   },
   "source": [
    "### Criando a feature marca"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "id": "TwQOwSUnnoC_"
   },
   "outputs": [],
   "source": [
    "#Criando a coluna marca\n",
    "df1['brand'] = df1['name'].str.split(' ').str[0]\n",
    "\n",
    "#para a base de teste\n",
    "df_test['brand'] = df_test['name'].str.split(' ').str[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "P7fQUSM60PJP",
    "outputId": "65ac56c1-2770-4116-f349-ca9bda84668c"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['Maruti', 'BMW', 'Tata', 'Toyota', 'Hyundai', 'Chevrolet', 'Honda',\n",
       "       'Jaguar', 'Renault', 'Mahindra', 'Volkswagen', 'Ford', 'Skoda',\n",
       "       'Datsun', 'Fiat', 'Volvo', 'Nissan', 'Mercedes-Benz', 'Kia',\n",
       "       'Jeep', 'Audi', 'Isuzu', 'Lexus', 'Land', 'Force', 'Mitsubishi',\n",
       "       'Ambassador', 'Daewoo', 'MG', 'Ashok'], dtype=object)"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df1['brand'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "TX_PMhJb0LBo",
    "outputId": "cbd5435f-44f9-44d4-e3fb-6637494942c6"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['Tata', 'Maruti', 'Mahindra', 'Hyundai', 'Volvo', 'Jaguar',\n",
       "       'Chevrolet', 'Jeep', 'Honda', 'Toyota', 'Kia', 'Ford', 'Lexus',\n",
       "       'Skoda', 'BMW', 'Fiat', 'Renault', 'Nissan', 'Datsun',\n",
       "       'Mercedes-Benz', 'Volkswagen', 'Opel', 'Mitsubishi', 'Ambassador',\n",
       "       'Audi', 'Land', 'Isuzu', 'Force'], dtype=object)"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_test['brand'].unique()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "KVJN1Gybqnn-"
   },
   "source": [
    "### Criando a coluna idade do carro"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "id": "5zSaXZGGoBrr"
   },
   "outputs": [],
   "source": [
    "#Criando a coluna idade do carro\n",
    "df1['age'] = 2021 - df1.year\n",
    "\n",
    "#para a base de teste\n",
    "df_test['age'] = 2021 - df_test.year"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "3_6Au6o-q_U7"
   },
   "source": [
    "### Retirando a coluna name e year"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "id": "hNozwzdSogMy"
   },
   "outputs": [],
   "source": [
    "del df1[\"name\"] #retirando a coluna name\n",
    "del df1[\"year\"] #retirando a coluna year\n",
    "del df1[\"torque\"] #retirando a coluna year ## 0.9705529686331869 (sem torque) gradiente\n",
    "#del df1[\"seats\"] #retirando a coluna year ## 0.9719490890165687 (sem seats e torque) gradiente\n",
    "#del df1[\"engine\"] #retirando a coluna year ### 0.9723490022839645 (sem engine, seats e torque) gradiente\n",
    "\n",
    "\n",
    "del df_test[\"name\"] #retirando a coluna name\n",
    "del df_test[\"year\"] #retirando a coluna year\n",
    "del df_test[\"torque\"] #retirando a coluna torque\n",
    "#del df_test[\"seats\"] #retirando a coluna year\n",
    "#del df_test[\"engine\"] #retirando a coluna year"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "bTGXKQpVrKB2"
   },
   "source": [
    "### Adequando o tipo de dado de algumas variáveis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "id": "TGUv4w3NrE_T"
   },
   "outputs": [],
   "source": [
    "#Mudando o tipo de dado de algumas variáveis\n",
    "df1['mileage'] = pd.to_numeric(df1['mileage'])\n",
    "df1['max_power'] = pd.to_numeric(df1['max_power'])\n",
    "df1['engine'] = pd.to_numeric(df1['engine']) \n",
    "df1['seats'] = pd.to_numeric(df1['seats']) \n",
    "#df1['torque'] = pd.to_numeric(df1['torque'])  \n",
    "\n",
    "#Base de Teste\n",
    "df_test['mileage'] = pd.to_numeric(df_test['mileage'])\n",
    "df_test['max_power'] = pd.to_numeric(df_test['max_power'])\n",
    "df_test['engine'] = pd.to_numeric(df_test['engine'])\n",
    "df_test['seats'] = pd.to_numeric(df_test['seats']) \n",
    "#df_test['torque'] = pd.to_numeric(df_test['torque']) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "aHiZuHmTz4AS"
   },
   "source": [
    "## Transformar as variaveis categoricas para a regressão"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "PA4-FryE4wwt"
   },
   "source": [
    "### Substituicao variaveis categoricas por rótulos numéricos\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "id": "0Ee4glR8IbfB"
   },
   "outputs": [],
   "source": [
    "from sklearn import preprocessing\n",
    "le = preprocessing.LabelEncoder()\n",
    "#base de treino\n",
    "for i in range(0, len(df1.columns.values)):\n",
    "  if df1.dtypes[i] == 'O':\n",
    "    df1.iloc[:, i] = le.fit_transform(df1.iloc[:, i]).astype('str')\n",
    "\n",
    "#Na base de test\n",
    "for i in range(0, len(df_test.columns.values)):\n",
    "  if df_test.dtypes[i] == 'O':\n",
    "    df_test.iloc[:, i] = le.fit_transform(df_test.iloc[:, i]).astype('str')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "wLSbwCkY8IuW",
    "outputId": "fc85f43b-276c-43ee-bc1b-85d831c8b5b4"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 5530 entries, 1 to 5688\n",
      "Data columns (total 12 columns):\n",
      " #   Column         Non-Null Count  Dtype  \n",
      "---  ------         --------------  -----  \n",
      " 0   selling_price  5530 non-null   int64  \n",
      " 1   km_driven      5530 non-null   int64  \n",
      " 2   fuel           5530 non-null   object \n",
      " 3   seller_type    5530 non-null   object \n",
      " 4   transmission   5530 non-null   object \n",
      " 5   owner          5530 non-null   object \n",
      " 6   mileage        5530 non-null   float64\n",
      " 7   engine         5530 non-null   int64  \n",
      " 8   max_power      5530 non-null   float64\n",
      " 9   seats          5530 non-null   float64\n",
      " 10  brand          5530 non-null   object \n",
      " 11  age            5530 non-null   int64  \n",
      "dtypes: float64(3), int64(4), object(5)\n",
      "memory usage: 561.6+ KB\n"
     ]
    }
   ],
   "source": [
    "df1.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 204
    },
    "id": "hKDzE8yj9I3N",
    "outputId": "5e844c57-e8a8-47b4-d118-6da1238d7e24"
   },
   "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>selling_price</th>\n",
       "      <th>km_driven</th>\n",
       "      <th>fuel</th>\n",
       "      <th>seller_type</th>\n",
       "      <th>transmission</th>\n",
       "      <th>owner</th>\n",
       "      <th>mileage</th>\n",
       "      <th>engine</th>\n",
       "      <th>max_power</th>\n",
       "      <th>seats</th>\n",
       "      <th>brand</th>\n",
       "      <th>age</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>459999</td>\n",
       "      <td>87000</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>20.77</td>\n",
       "      <td>1248</td>\n",
       "      <td>88.76</td>\n",
       "      <td>7.0</td>\n",
       "      <td>20</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1100000</td>\n",
       "      <td>102000</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>19.62</td>\n",
       "      <td>1995</td>\n",
       "      <td>187.74</td>\n",
       "      <td>5.0</td>\n",
       "      <td>3</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>229999</td>\n",
       "      <td>212000</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>11.57</td>\n",
       "      <td>2179</td>\n",
       "      <td>138.10</td>\n",
       "      <td>7.0</td>\n",
       "      <td>26</td>\n",
       "      <td>12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>800000</td>\n",
       "      <td>125000</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>11.50</td>\n",
       "      <td>2982</td>\n",
       "      <td>171.00</td>\n",
       "      <td>7.0</td>\n",
       "      <td>27</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>180000</td>\n",
       "      <td>25000</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>19.70</td>\n",
       "      <td>796</td>\n",
       "      <td>46.30</td>\n",
       "      <td>5.0</td>\n",
       "      <td>20</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   selling_price  km_driven fuel seller_type transmission owner  mileage  \\\n",
       "1         459999      87000    1           1            1     0    20.77   \n",
       "2        1100000     102000    1           0            0     0    19.62   \n",
       "3         229999     212000    1           1            1     4    11.57   \n",
       "4         800000     125000    1           1            1     2    11.50   \n",
       "5         180000      25000    3           1            1     2    19.70   \n",
       "\n",
       "   engine  max_power  seats brand  age  \n",
       "1    1248      88.76    7.0    20    9  \n",
       "2    1995     187.74    5.0     3   11  \n",
       "3    2179     138.10    7.0    26   12  \n",
       "4    2982     171.00    7.0    27   11  \n",
       "5     796      46.30    5.0    20   11  "
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df1.head(5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "R0QjbNF25Ok-"
   },
   "source": [
    "Temos agora todas as variaveis numericas para utilizar na nossa regressao"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 407
    },
    "id": "v_gj0KH_5R8T",
    "outputId": "06b9f4ed-7c7d-4999-c796-7d241102f83e"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x432 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(16, 6))\n",
    "# define the mask to set the values in the upper triangle to True\n",
    "mask = np.triu(np.ones_like(df1.corr(), dtype=bool))\n",
    "heatmap = sns.heatmap(df1.corr(), mask=mask, vmin=-1, vmax=1, annot=True, cmap='BrBG')\n",
    "heatmap.set_title('Correlation Heatmap', fontdict={'fontsize':18}, pad=16);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "We1FtfYS6cmW"
   },
   "source": [
    "Correlacoes muito altas entre variaveis do treino poderiam ser reduntantes, observamos que isso nao ocorre"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "VUI9wNF-t52a"
   },
   "source": [
    "### Analise das variaveis quantitativas\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "id": "VUSsXDLnEK9R",
    "outputId": "a8c66c0e-c57e-4fe0-bba4-f9a22f92c988"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYcAAAEmCAYAAACJXlw1AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8QVMy6AAAACXBIWXMAAAsTAAALEwEAmpwYAAAxvUlEQVR4nO3de1wU5f4H8M8CAqIZaVxKFLEEFUmUOpW3QhFQQQRveMPUvISXDueYoKBUpJJReky7mJ4stWNmImaG17QM83YsU0FMBJVcQFARubPP7w9/O4d1lmVRYUf5vF8vXy9m5tmZz+z68N258IxKCCFARERUjZmpAxARkfKwOBARkQyLAxERybA4EBGRDIsDERHJsDgQEZEMi0MjcfnyZXTq1AlBQUEICgpCYGAghg8fjuPHj9fL9tzc3FBQUGCwzf79+/Gvf/3rnrYTFRWFNWvW1Lj9P/74A7NmzTK4jpMnT2LBggX3lEMp1qxZg6ioqDq9pqb38H4rKCiAm5ub3mV79+7FO++8U+8ZyHgWpg5ADcfa2hpJSUnS9I4dOzB37lzs2rXLJHn++OMP3Lhxo1634eHhgeXLlxts8+effyInJ6dec5Bh/fr1Q79+/Uwdg6phcWjErl+/Djs7O2n666+/xrp162BmZobHH38c8+fPh7OzMyZMmAB3d3fMmTMHKSkpiIqKwpYtW5CQkAArKyukpaUhPz8fPXv2RExMDJo0aaKznZUrV+L777+Hubk5XFxcMH/+fPz111/YuHEjqqqq8MgjjyAiIkLnNefPn8fChQtx/fp1VFVVYdy4cRg2bFid9/Hw4cOIi4vD9u3bcezYMcTHx0Oj0QAApk6dimeeeQbLly/HzZs3MXfuXCxevFjv++Di4oKCggLMnTsXFy9ehK2tLezs7NChQwfMnDkTXbp0Qb9+/ZCWloaEhAScPXsWX3/9NSoqKnDjxg1MnjwZo0ePxpYtW7Br1y5oNBr89ddfcHBwwIgRI7B+/XpkZmZiwoQJmDhxIoqLi/Hmm28iKysL169fR7NmzZCQkID27dvr7F9FRQXeeecdpKSkoFWrVmjVqhUeeeQRAMDNmzexcOFCpKeno6KiAi+++CLmzJkDC4uau/2iRYtw9uxZfPTRR4iLi4O1tTXS09ORn5+Pvn37wtbWFj/++CPy8vLwzjvv4MUXXzT4/u/atQtLly5F06ZN0aVLF2n+li1bsHnzZpSUlKB58+YIDg7Gzp07ERUVhdDQUPz888+wtLREVVUVXn75Zaxduxb29vY17o+HhwemTJmCX375Bbm5uXj11VcxevToOv9/oWoENQqXLl0SHTt2FIMHDxaDBw8WL7/8snB3dxf79+8XQgiRkpIifHx8RH5+vhBCiG+//VYMGDBAaDQakZOTI3r06CF2794tevfuLY4cOSKEECIyMlIMGTJEFBUVibKyMjFmzBixbt06IYQQrq6uIj8/X2zevFmMHDlS3Lp1SwghxPLly8XEiROln9966y1Z1oqKCjFw4EBx6tQpIYQQhYWFYsCAAeLEiROytpGRkaJXr17Sfmn/abf/66+/ikGDBgkhhAgLCxPbt28XQgiRmpoq3nzzTWlfp0yZUuv7EBERIZYsWSKEECInJ0f07NlTLF++XNrfxMREIYQQRUVFYsSIEaKgoEAIIcSJEyeEp6entD4vLy/x119/iaqqKjFw4EAxc+ZMUVVVJVJTU4WHh4eoqqoSP/zwg4iLi5P2c/78+eLtt9+W7f/atWtFWFiYKCsrE7du3RLBwcEiMjJSCCFEVFSU+PLLL4UQQlRWVorZs2eLVatW6X0PP/vsM/HWW2+J6dOni7KyMmn+8OHDRXl5ucjNzRWurq7S+tauXSsmTJggW1d1eXl5wsvLS5w7d04IIcQnn3wiXF1dpffhueeeEzdv3pR9BmPGjBE//PCDEEKI/fv3i9DQ0Fr3x9XVVfq/98cff4guXbqI0tJSg/nIMB45NCJ3nlZKSUnB9OnTsW3bNvz8888YOHAgWrZsCQAICQnBwoULcfnyZbRp0wZxcXEIDw/HzJkz8dxzz0nrCA4ORrNmzQAAQUFB2Lt3L8aOHSst/+mnnxASEgIbGxsAQFhYGD755BOUl5fXmDMzMxMXL17EvHnzpHmlpaU4c+YMPD09Ze1feeUVTJo0SWeevnPbAwYMwNtvv419+/ahR48e+Mc//iFrY+h9OHDgABITEwEA9vb28Pf313nts88+CwBo1qwZPvnkExw4cACZmZlIS0tDcXGx1M7DwwNPPPEEAMDJyQm9evWCmZkZ2rRpg7KyMpSUlMDf3x9t2rTBunXrkJWVhSNHjqBbt26yvIcOHUJAQAAsLS1haWmJwMBAnD17FsDtazp//PEHNm/eLL2HNVm7di3y8/OxdetWWFpaSvO9vb3RpEkT2NnZwcbGBr179wYAtG3bFtevX69xfQBw/PhxuLq64umnnwYAjBw5Eh988IG03M3NDc2bN5e9btiwYUhMTIS/vz+2bNmCESNGGLU/2tNS7u7uKC8vR3FxMaysrAxmpJqxODRiPXr0QNu2bfHHH39Ip1qqE0KgsrISwO3z8o8//jhOnjyp08bc3FynvZmZ7j0OGo0GKpVKZ1q7zppoTzVVL2RXr16VTpfcrdDQUHh7e+OXX37Bzz//jBUrViA5OVmW907a98HCwgKi2lBkd+6rtgCq1WqMHDkSI0aMgJeXF/z9/fHjjz9K7ar/8gWg9zTPV199hU2bNmHMmDEIDAyEra0tLl++XOs+Vv88NBoN/vWvf+Gpp54CABQWFup8FtU999xz6N69O+bOnYuvv/5aOjVoTFZDqr9fd75W+37dacCAAYiPj8f58+dx9OhRxMfHG7U/2kKgnSc4bNw94d1KjdiFCxeQnZ2NTp06oXfv3tixY4d0h9G3334LW1tbODs74+TJk/jyyy/x7bff4ubNm/jiiy+kdfzwww8oLy9HWVkZEhMT4e3trbON3r1749tvv5W+Oa9btw7PPfccLC0tYW5urrdQuLi46BzlXLlyBQEBATh16tQ97W9oaChSU1MREhKCuLg4FBYWIi8vTyeHoffhpZdekr61Xrt2DXv27NH7y/bUqVNo2bIlwsPD0atXL6kwVFVVGZ314MGDCA4OxvDhw+Hi4oJ9+/bpfX3v3r2xdetWlJWVoaysDDt27JCW9erVC2vXroUQAuXl5Xjttdewfv16vdvr0qULxo4di0ceeQQrVqwwOqchzz33HP7880+kpaUBuH2dwRhWVlYYNGgQoqKi4Ovri6ZNm9Z5f+je8cihESktLUVQUJA0rdFo8Pbbb8PFxQUuLi545ZVXMH78eGg0GrRs2RKffvopiouL8Y9//AMxMTFwcHBAfHw8hg8fLp1asra2xujRo1FYWAg/Pz8MHTpUZ5vDhg3DlStXMHz4cGg0Gjg7OyMhIQEA8MILL2D27NmIi4vD/PnzpddYWlrio48+wsKFC7F69WpUVlbi9ddfh5eX1z3t/+zZs7Fo0SIsW7YMKpUKM2bMgJOTE6qqqrBy5UrMmDEDK1as0Ps+mJmZYe7cuYiJiZG+yT/55JOwtraWbadnz57YvHkz/P39oVKp8Le//Q0tW7ZEVlaW0VknTpyIBQsWSMXI09MT6enpsnahoaG4ePEiAgICpCKmFR0djYULFyIwMBAVFRXo0aMHXn311Rq3qVKpsGjRIgwZMgQvvfSS0Vlr0rJlSyQkJGD27Nlo0qSJzunI2gwfPhzr16/Hm2++Kc2r6/7QvVEJHnvRXYqKikKHDh1k5/sfVhs2bEDnzp3RrVs3lJeXY/To0Zg5c+Z9+UVKpDQ8ciAy0tNPP424uDhoNBpUVFTA39+/0ReG1atX47vvvtO7bNKkSRg8eHADJ6L7hUcOREQkwwvSREQkw+JAREQyLA5ERCTD4kAm8/3332PAgAHw9PSEj48Pjh07Ji07f/48wsLC4OXlhf79+2P37t0AgPLycsybNw/e3t7o1q0bhgwZggMHDhi97m7duun869SpE+Li4mrMuH79eoSEhKBLly6y0U6NzdKQxo0bBw8PD2n//Pz8dJYb2h+tzMxMeHh4YPbs2Trzr1+/junTp8PT0xPe3t6yC9HGrFurps+3OkP/P6gBmGbUDmrsDh48KF5++WVx4sQJUVVVJdRqtVCr1UKI22Mr+fr6in//+9+isrJSpKSkiK5du4qMjAxx69YtsXz5cnHp0iVRVVUl9u3bJzw9PcWlS5eMWnd1t27dEp6entJYUfrs3LlT7N69WyxYsEAas6j662vL0tDGjh0rNm3aVONyQ/ujNWHCBDFq1Cjxz3/+U2d+RESEeP3110VRUZE4evSo6N69u0hPT6/TuoUw/PlqGfsZUv3hkQMZ9PHHHyM2NlaavnHjBtzd3VFWVnZP6/3www8RHh4OT09PmJmZwcHBAQ4ODgCAjIwM5Obm4pVXXoG5uTlefPFFdO/eHUlJSbCxscHMmTPh5OQEMzMzeHt7w8nJCadPnzZq3dXt3LkTLVu2lMZE0sfX1xc+Pj6wtbWVLTMmS00qKiqwdOlS9O3bF+7u7nBzc4Obm1u93/ppaH+A29/WH3nkEdloq8XFxdi1axdef/11NGvWDM8++yz69u2rM8RJbevWMvT5ahn7GVL94d85kEHp6el4/vnnpenU1FS4uLjIBjSbOnVqjQ8O8vLywqeffipNV1VV4dSpU+jbty/69++PsrIy+Pj4YM6cObC2ttY7Jo4QAufOnZPNv3r1KjIzM6XB3Wpbd3WJiYkYMmRIjeMN1dWdWQxZtmwZjh49ig0bNuDRRx9FeHg4mjdvjsjISJ12dXlftd5//30kJCTAxcUFEREROp+fIUVFRVi+fDnWrl0r/WW2VmZmJszMzODi4iLN69ixI44ePWrUuqur7fOty2dI9YfFgQxKT0/HK6+8Ik2npaXpHfFU3y+pmly9ehUVFRVITk7Ghg0bYGFhgfDwcHz88ceIiIhA+/bt0bJlS6xevRqvvPIKDh8+jKNHj8p+yVVUVGD27NkIDg6WBmOrbd1af/31F44ePYqFCxfW8R3RT1+WmhQVFWHdunXYtm2bNDqrr68vfvjhB7Rp00anbV3eV+D2ECFPPfUULC0t8f3332PatGlISkpC27Zta33tsmXLMHToUClTdcXFxbKBDx955BHcunWrTvkA1Pr5GvsZUv3iaSWqUXl5OS5evAhXV1dpXlpaGjp16nRP69V++xs3bhzs7e3RsmVLTJgwQbqY26RJE6xcuRIHDhxAr1698Pnnn8Pf31/ntIJGo8GcOXPQpEkTnXGZalu31tatW+Hl5SX7ZXw3aspSk2PHjqFNmzZo166dNK+wsBCPP/74PWfp2rUrmjdvDktLSwQHB6N79+5GXSRPTU3FoUOHdL4IVGdjY4OioiKdeUVFRdJw7XVR2+dr7GdI9YtHDlSj8+fPw8HBQRoVUwiBI0eOICAgQNb21VdfNXj6Y/Xq1dL0o48+CkdHR4Onczp27Kgz4mZoaCiGDBki5YiOjsbVq1fx2Wef6Tx5zph1A0BSUhImT55ssI0xDGWpSUFBAVq0aKGzjt27dyMsLEzWti7vqz4qlcqooasPHz6M7OxsaVTd4uJiVFVVITg4GImJiWjXrh2qqqqQmZkpFbW0tDSjTqHpY+jzNfYzpPrF4kA1Onv2LPLz83Hx4kXY29vj448/RnZ2Nlq3bi1rW9svqTuFhIRg3bp16N27NywsLPDFF1/g5ZdflpanpaXBxcUFGo0GX331FXJzcxESEgIAiI2Nxfnz5/H555/rPQdd27r/+9//IicnR/awHn0qKytRVVUFjUaDqqoqlJWVwdzcXHo2QW1Z9OnQoQPOnDkjXb9ZsWIFVCoVBg4cKGtbl/e1sLAQv//+O/72t7/B3NwcO3bswLFjx3QemlTT/owcORKDBg2S2v373/9Gdna2NCqqjY0N+vfvj+XLl+Odd95Bamoq9u7di40bNxr9XlVn6PMFav8MqQGY7D4pUrx3331XzJw5U/j6+opevXqJL7/8UvTr10/MmTPnntddXl4uYmNjhZeXl+jRo4eIi4vTeaxjfHy8ePbZZ4Wnp6eYNGmSyMzMFEIIcfnyZeHq6iq6dOkiPD09pX9JSUlGr3v+/Pli9uzZenNNmjRJfPzxx9L08uXLhaurq84/7aNBa8vy6quvij179ujdzkcffSR69uwpevbsKSIjI6XHkt6L/Px8ERISIjw9PYWXl5cYPny4OHjwoE4bQ/tzZ7s7b2W9du2aeO2110TXrl3FSy+9JLZt22b0uu98X2v6fLVq+wyp/nHgParRq6++iuHDh8v+kIqMs2nTJjg6OqJPnz6mjkJUZ7wgTTVKT0+v9c4bqpn2Hn6iBxGPHEivGzduoGfPnjhx4oRRF1mJ6OHC4kBERDI8rURERDIsDkREJMPiQEREMiwOREQkw+JAREQyLA5ERCTD4kBERDIsDkREJMPiQEREMiwOREQkw+JAREQyfNhPDT777DNkZGTozLt27RoA4LHHHqvxde3bt78vTxgjIjIlFocaZGRk4NSZszC3tpXmVZVeBwCor5XrfY12ORHRg47FwQBza1vYOPeTpouz9gKAzrzqtMuJiB50vOZAREQyLA5ERCTD4kBERDIsDkREJMPiQEREMiwOREQkw+JAREQyLA5ERCTD4kBERDIsDkREJMPiQEREMiwOREQkw+JAREQyLA5ERCTD4kBERDIsDkREJMPiQEREMo26OOzbtw/79u0zdQwZpeYiosajUT8mdPfu3QCAvn37mjiJLqXmIqLGo1EfORARkX4sDkREJMPiQEREMiwOREQkw+JAREQyLA5ERCTD4kBERDIsDkREJMPiQEREMiwOREQkw+JAREQyLA5ERCTD4kBERDIsDkREJNOoh+xWqmvXriE7OxuBgYG1tu3Tpw9++uknODs7IysrC08//TT+/PNPaXlkZCS2bt2KkpIS5ObmokWLFsjNzYWTkxMWLVoEIQSWLFmC0NBQLF68GPHx8XBxcZFtp6CgAAsXLkRZWRlyc3Px7rvv6m1Xvf2SJUsQGRkpbSMyMhKPPfZYje20y/TNIyK5+uwrPHJQoOzsbKPb/vTTTwCArKwsANApDADw/vvv4+zZs7h48SJKS0uRm5sLALh8+TI2btyIjRs34syZM3j33XdRXFyMhIQEvdvZuHEj0tPTkZWVhZKSkhrbVW9/5swZnW1s3LjRYDtD84hIrj77CouDwmh/2d8vlZWVNS5LTk7Gnj17IIRAUVERAODixYu4cOGCTruCggLs2bNHZ56+dtXb7927F0II7N69W9rGnj17cO3aNb3ttMv0zSMiufruK436tNL169dRUFCAuXPnypZlZGRAU2lep/VpKkuRkZGhd33GOnXq1F2/tq40Gg2EELL5CQkJWLlypTS9ceNGvUXmznbV22s0GgC6xUmj0WDjxo147bXXZO20y4QQsnna9kT0P/r6z/3sKzxyaOT0FYeLFy/qTO/fv9+odtXba4uCEEJ6bWVlJX788Ue97bTL9M0jIrn67iuN+sjB1tYWtra2WLx4sWzZ3LlzkZqRU6f1mVlYo317B73rM1ZwcLDBU0H3m0qlkv3ib9u2rc70yy+/jOTk5FrbVW+/e/duVFZWQqVSAbhdJCwsLODt7a23nXaZ9lRU9XlEJKev/9xPPHJQmIiIiAbblpmZGSws5N8PZs+erTMdGhpqVLvq7c3Mbv/XsrCwkF5rZmaG0NBQve20y/TNIyK5+u4rLA4K06dPn/u6Pn2/1LX8/f3h4+MDlUqF5s2bA7h9NHDnLaotW7aEj4+Pzjx97aq379evH1QqFfr37y9tw8fHR+d2u+rttMv0zSMiufruKywOCtS6dWuj22qLibOzMwDg6aef1ln+z3/+E25ubmjbti2sra1hb28PAHBycpK+qXfu3BmRkZGwsbExeDTg6uoKZ2dnNG3atMZ21dt37txZZxv6vtnoW2aoPRH9T332FZXQd6WxkdDeVWTomoONcz9pXnHWXgDQmVddcdZedLrHaw615SIiagg8ciAiIhkWByIikmFxICIiGRYHIiKSYXEgIiIZFgciIpJhcSAiIhkWByIikmFxICIiGRYHIiKSYXEgIiIZFgciIpJhcSAiIhkWByIikmnUjwnt37+/qSPopdRcRNR4NOri0LdvX1NH0EupuYio8eBpJSIikmFxICIiGRYHIiKSYXEgIiIZFgciIpJhcSAiIhkWByIikmFxICIiGRYHIiKSYXEgIiIZFgciIpJhcSAiIhkWByIikmFxICIiGRYHIiKSYXEgIiIZFgciIpJp1E+Cq01V6XUUZ+3VmQagM+/O9oBD/QcjIqpnLA41aN++vWzetWuWAIDHHnushlc56H0dEdGDRiWEEKYOQUREysJrDkREJMPiQEREMiwOREQkw+JAREQyLA5ERCTD4kBERDIsDkREJPPA/xFcZWUl1Gq1qWMQET2QHB0dYWEhLwUPfHFQq9Xo16+fqWMQET2Q9u7dCycnJ9n8B/4vpO/lyEGtVmPMmDHYsGEDHB0d73Oy+0PpGZWeD2DG+0Hp+QDlZ1Rqvof2yMHCwkJv1asLR0fHe15HfVN6RqXnA5jxflB6PkD5GZWeT4sXpImISIbFgYiIZFgciIhIplEXhxYtWmDGjBlo0aKFqaPUSOkZlZ4PYMb7Qen5AOVnVHq+Oz3wdysREdH916iPHIiISD8WByIikmk0xeG7777DwIED4evriw0bNsiWp6amIiQkBH5+foiOjkZlZaXiMu7ZswdBQUEYPHgwwsPDcePGDUXl09q/fz/69u3bgMn+p7aMGRkZGDduHAYPHoxJkyYp7j08ffo0hg4disGDB2Pq1KkoLCxs0HxaRUVFCAgIwOXLl2XLlNBXDOUzdT/RMpRRy5R9pVaiEVCr1cLb21tcu3ZN3Lp1SwQGBopz587ptBk0aJA4ceKEEEKIuXPnig0bNigq482bN0XPnj2FWq0WQgixbNkyERcXp5h8Wnl5ecLf3194e3s3WDZjM2o0GuHr6ysOHDgghBDivffeE0uWLFFMPiGEGDVqlNi/f78QQojFixeLDz74oMHyaf32228iICBAuLu7i0uXLsmWm7qvGMpn6n5iTEYtU/YVYzSKI4eUlBS88MILsLW1hY2NDfz8/JCcnCwtz87ORmlpKTw9PQEAISEhOsuVkLGiogKxsbFwcHAAALi5ueHKlSuKyacVExODGTNmNFiu6mrLePr0adjY2KBPnz4AgGnTpmHMmDGKyQcAGo0Gt27dAgCUlJTA2tq6wfJpbdq0CbGxsbC3t5ctU0JfMZTP1P1Ey1BGLVP2FWM88MNnGCM3Nxd2dnbStL29PU6ePFnjcjs7O+Tk5Cgq42OPPYb+/fsDAEpLS7Fq1SqMGzdOMfkA4Msvv0Tnzp3RtWvXBstVXW0ZL168iMcffxzz5s1Damoq2rdvj/nz5ysmHwBERUVh4sSJWLRoEZo2bYpNmzY1WD6thQsX1rhMCX3FUD5T9xMtQxkB0/cVYzSKIweNRgOVSiVNCyF0pmtbroSMWjdv3sSUKVPQsWNHBAcHKyZfeno6du3ahfDw8AbLdKfaMlZWVuLIkSMYNWoUEhMT0aZNG8THxysmX2lpKaKjo7F27VocPHgQo0ePRmRkZIPlM4YS+ooxTNVPjKGEvmKMRlEcHB0dkZeXJ03n5eXpHO7dufzq1asGDwdNkRG4/a1t9OjRcHNzq/WbSUPnS05ORl5eHoYOHYopU6ZIWZWU0c7ODs7OzvDw8AAABAQEyL65mzJfeno6rKys8MwzzwAARo4ciSNHjjRYPmMooa/UxpT9xBhK6CvGaBTFoUePHjh06BAKCgpQUlKCXbt2SeedAaB169awsrLC8ePHAQBJSUk6y5WQsaqqCtOmTcOAAQMQHR3d4N/Wass3a9Ys7Ny5E0lJSVi1ahXs7e3x1VdfKSpjt27dUFBQgLS0NADAvn374O7urph8zs7OUKvVyMjIAHB7nH1tIVMKJfQVQ0zdT4yhhL5ijEZxzcHBwQEREREICwtDRUUFhg0bhmeeeQaTJ0/GrFmz4OHhgYSEBMTExKCoqAju7u4ICwtTVEa1Wo0zZ86gqqoKO3fuBAB06dKlwb4ZGfMempoxGVeuXImYmBiUlJTA0dERS5YsUVS+xYsX4+9//zuEEGjVqhUWLVrUYPkMUVJfMZTP1P3EECX1FWM88MNnaB/2U9MDK4iIqO4e+NNK2seE8jnSRET3zwNfHIiI6P5jcSAiIhkWByIikmFxICIiGRYHAxLWH0fC+uOmjkFE1OB476cBJWUNPxQxEZES8MihDngkQUSNBY8c6qD6kYS2SMwe62WqOERE9YbF4S7xlBMRPcx4WomIiGRYHIiISIbFgYiIZOq9OLz77ruIiooCcPsZuoGBgfD19cXSpUulNqmpqQgJCYGfnx+io6NRWcnz+UREplSvxeHQoUNITEwEcPsRiPPmzcNHH32EHTt24NSpUzhw4AAA4I033sCCBQuwc+dOCCFM8txcIiL6n3orDtevX8fSpUsxbdo0AMDJkyfh7OyMNm3awMLCAoGBgUhOTkZ2djZKS0vh6ekJAAgJCUFycrLedRYWFuLy5cs6/xpqqO7VW0/J5llbmmPlN783yPaJiBpSvd3KumDBAkRERODKlSsAbj/X1c7OTlpub2+PnJwc2Xw7Ozvk5OToXecXX3yBFStW1Fdkg0rL9Z/q4i2tRPQwqpfi8M033+CJJ57Aiy++iC1btgAANBqNzvNchRBQqVQ1ztdn/PjxCA4O1pmnVqsxZsyYetgLIqLGq16Kw44dO5CXl4egoCDcuHEDxcXFyM7Ohrm5udQmLy8P9vb2cHR0RF5enjT/6tWrsLe317veFi1aoEWLFvURmYiIqqmX4vD5559LP2/ZsgVHjhzBW2+9BV9fX2RlZcHJyQnbt2/H0KFD0bp1a1hZWeH48ePw8vJCUlIS+vTpUx+xiIjISA32dw5WVlaIj4/HzJkzMXDgQLRv3x7+/v4AgISEBCxevBj+/v4oLi5GWFhYQ8Wqd7xgTUQPonofWykkJAQhISEAgBdffBHbtm2TtenYsSM2b95c31FMghesiehBxL+QJiIiGRYHIiKSYXEgIiIZFgciIpJhcSAiIhkWh2p42ykR0W0sDtXwtlMiottYHIiISIbFgYiIZFgciIhIhsWBiIhkWByIiEiGxYGIiGRYHIiISMao4jBv3jzZvFmzZt33MEREpAwGn+cQGxuLnJwcHD9+HAUFBdL8yspKXLp0qd7DERGRaRgsDsOGDcO5c+dw9uxZ+Pn5SfPNzc3h6elZ39mIiMhEDBYHDw8PeHh4oEePHnB0dGyoTCaxeuspU0cgIlIMox4TeuXKFbzxxhu4ceMGhBDS/O+++67egjW00nLdcZVYLIioMTOqOCxYsAAhISHo3LkzVCpVfWdShDuLBRFRY2JUcbCwsMCECRPqO8sDLWH9cQDA7LFeJk5CRHTvjLqVtUOHDjh79mx9Z3mglZRVQgjBZ0IQ0UPBqCOHS5cuYejQoXjyySdhZWUlzX+YrjkAgLWlORLWH4e1pfldr4PPhCCih4FRxSEiIqK+cyiG9gigqZVRbw0R0UPJqN+Arq6u9Z2DiIgUxKji8MILL0ClUkEIId2tZGdnh59++qlewxERkWkYVRzS0tKkn8vLy7F9+3ZcuHCh3kIREZFp1XlUVktLS4SEhOCXX36pjzwPPO1Fbe2trUREDyKjjhyuX78u/SyEwKlTp1BYWFhfmR54vGOJiB50db7mAACtWrVCdHR0vQYjIiLTqfM1ByIievgZVRw0Gg3WrFmDn376CZWVlejZsyemTZsGCwv+LQAR0cPIqAvS77//Pn799VeMHz8eEyZMwIkTJ7BkyZJaX7dixQoMGjQIgwYNktqnpKQgMDAQvr6+WLp0qdQ2NTUVISEh8PPzQ3R0NCored6eiMhUjCoOP//8Mz755BP4+PjA19cXH3/8ca1/45CSkoKDBw8iMTERW7duxenTp7F9+3bMmzcPH330EXbs2IFTp07hwIEDAIA33ngDCxYswM6dOyGEwKZNm+5974iI6K4YVRyEEGjSpIk0bWlpqTOtj52dHaKioqS2Tz31FDIzM+Hs7Iw2bdrAwsICgYGBSE5ORnZ2NkpLS6Wny4WEhCA5Ofnu94qIiO6JURcNOnbsiEWLFmHs2LFQqVRYt25drUNqdOjQQfo5MzMTP/zwA8aOHQs7Oztpvr29PXJycpCbm6sz387ODjk5ObJ1FhYWym6hVavVxuwCERHVgVHFITY2Fu+88w5CQ0Oh0WjQu3dvzJ8/36gNnDt3DlOnTsWcOXNgbm6OzMxMaZl2OA6NRqPzEKHqw3RU98UXX2DFihVGbZeIiO6eweJQXl6O+fPnw8fHB/Hx8QCAKVOmwNzcHM2bN6915cePH8esWbMwb948DBo0CEeOHEFeXp60PC8vD/b29nB0dNSZf/XqVdjb28vWN378eAQHB+vMU6vVGDNmTK1ZiIjIeAavOSxfvhxFRUXo3r27NC8uLg6FhYX48MMPDa74ypUrmD59OhISEjBo0CAAQNeuXXHhwgVkZWWhqqoK27dvR58+fdC6dWtYWVnh+PHbQ04kJSWhT58+snW2aNECTk5OOv8cHR3rvNNERGSYwSOH/fv3Y/PmzbC2tpbmOTg4YMmSJRg5cqTB5zysWbMGZWVl0hEHAISGhiI+Ph4zZ85EWVkZXnrpJfj7+wMAEhISEBMTg6KiIri7uyMsLOxe942IiO6SweLQpEkTncKg1bx5c1haWhpccUxMDGJiYvQu27Ztm2xex44dsXnzZoPrJCKihmHwtJKZmRmKiopk84uKivhHakREDzGDxSEgIAAxMTEoLi6W5hUXFyMmJga+vr71Hk4JrC3NsXrrKVPHICJqUAZPK40fPx6xsbHo2bMnOnToAI1Gg/PnzyMwMBDTp09vqIz1KmH9cVhbmhtsU1rOoyQialwMFgczMzPExcVh2rRpOH36NMzMzPDMM8/ovc30QVVSVgkhBJpacRBBIiIto34jtm7dGq1bt67vLEREpBB1fkwoERE9/Fgc6ogXqImoMWBxuAu8QE1EDzsWByIikmFxICIiGRYHIiKSYXEgIiIZFgciIpJhcSAiIhkWByIikmFxICIiGRYHIiKSYXEgIiIZFgciIpJhcSAiIhkWByIikmFxICIiGRYHIiKSYXG4R3z4DxE9jFgc7gM+/IeIHjYWpg7Q2CSsPw5rS3OoVCpMH97V1HGIiPRicahn1pbmSFh/HAAwe6wXSsoqIYSASqUycTIiopqxODSAkjKediKiBwuvORARkQyLAxERyfC0Uj258xbXmm555QVqIlIiHjnUoztvcdV3y2tJWaX0ryYrv/n9vmcjIjKExcFE9B1ZJKw/Lt3ZVL0g8II2ETU0RRWH7777DgMHDoSvry82bNhg6jj17s4jCe1trqu3nmJBICKTUsw1h5ycHCxduhRbtmyBpaUlQkND8fzzz+Ppp582dbQGV1peKR1JWFua6/xs6NrEym9+53ULIrovFFMcUlJS8MILL8DW1hYA4Ofnh+TkZMyYMUNqU1hYiMLCQp3XZWdnAwDUavVdbbey5BoqKs1gVmHx/9Pl0nT1n02x7GZxOcqbmMGqiYX0s0qlwuXLl7FuRyoAYNzATv97f67l1rhMy9AyImp8HB0dYWEhLwUqIYQwQR6ZTz/9FMXFxYiIiAAAfPPNNzh58iTi4uKkNh9++CFWrFhhqohERA+dvXv3wsnJSTZfMUcOGo1GZ0gJfUNMjB8/HsHBwTrzysvLcenSJbRr1w7m5uZ12qZarcaYMWOwYcMGODo63n34eqT0jErPBzDj/aD0fIDyMyo1X01ZFFMcHB0dcezYMWk6Ly8P9vb2Om1atGiBFi1ayF7bvn37e962vsqpJErPqPR8ADPeD0rPByg/o9LzaSnmbqUePXrg0KFDKCgoQElJCXbt2oU+ffqYOhYRUaOkmCMHBwcHREREICwsDBUVFRg2bBieeeYZU8ciImqUFFMcACAwMBCBgYGmjkFE1Ogp5rSSKbRo0QIzZszQex1DKZSeUen5AGa8H5SeD1B+RqXnu5NibmUlIiLlaNRHDkREpB+LAxERyTSa4lDboH6pqakICQmBn58foqOjUVnZ8APf1ZZxz549CAoKwuDBgxEeHo4bN24oKp/W/v370bdv3wZM9j+1ZczIyMC4ceMwePBgTJo0SXHv4enTpzF06FAMHjwYU6dOlQ0X01CKiooQEBCAy5cvy5Ypoa8YymfqfqJlKKOWKftKrUQjoFarhbe3t7h27Zq4deuWCAwMFOfOndNpM2jQIHHixAkhhBBz584VGzZsUFTGmzdvip49ewq1Wi2EEGLZsmUiLi5OMfm08vLyhL+/v/D29m6wbMZm1Gg0wtfXVxw4cEAIIcR7770nlixZoph8QggxatQosX//fiGEEIsXLxYffPBBg+XT+u2330RAQIBwd3cXly5dki03dV8xlM/U/cSYjFqm7CvGaBRHDtUH9bOxsZEG9dPKzs5GaWkpPD09AQAhISE6y5WQsaKiArGxsXBwcAAAuLm54cqVK4rJpxUTE6MzWGJDqi3j6dOnYWNjI/1x5bRp0zBmzBjF5ANuDyNz69YtAEBJSQmsra0bLJ/Wpk2bEBsbKxuhAFBGXzGUz9T9RMtQRi1T9hVjKOrvHOpLbm4u7OzspGl7e3ucPHmyxuV2dnbIyclRVMbHHnsM/fv3BwCUlpZi1apVGDdunGLyAcCXX36Jzp07o2tX0wwbXlvGixcv4vHHH8e8efOQmpqK9u3bY/78+YrJBwBRUVGYOHEiFi1ahKZNm2LTpk0Nlk9r4cKFNS5TQl8xlM/U/UTLUEbA9H3FGI3iyKG2Qf2MGfTP1Bm1bt68iSlTpqBjx46yQQhNmS89PR27du1CeHh4g2W6U20ZKysrceTIEYwaNQqJiYlo06YN4uPjFZOvtLQU0dHRWLt2LQ4ePIjRo0cjMjKywfIZQwl9xRim6ifGUEJfMUajKA6Ojo7Iy8uTpu8c1O/O5VevXjV4OGiKjMDtb22jR4+Gm5tbrd9MGjpfcnIy8vLyMHToUEyZMkXKqqSMdnZ2cHZ2hoeHBwAgICBA9s3dlPnS09NhZWUlDRszcuRIHDlypMHyGUMJfaU2puwnxlBCXzFGoygOtQ3q17p1a1hZWeH48dvPb05KSmrwQf9qy1hVVYVp06ZhwIABiI6ObvBva7XlmzVrFnbu3ImkpCSsWrUK9vb2+OqrrxSVsVu3bigoKEBaWhoAYN++fXB3d1dMPmdnZ6jVamRkZAC4Pc6+tpAphRL6iiGm7ifGUEJfMUajuOZQ06B+kydPxqxZs+Dh4YGEhATExMSgqKgI7u7uCAsLU1RGtVqNM2fOoKqqCjt37gQAdOnSpcG+GRnzHpqaMRlXrlyJmJgYlJSUwNHREUuWLFFUvsWLF+Pvf/87hBBo1aoVFi1a1GD5DFFSXzGUz9T9xBAl9RVjcPgMIiKSaRSnlYiIqG5YHIiISIbFgYiIZFgciIhIhsWBiOgBZczgfsDdDTjJ4kBE9AD6/fffMWrUKGRmZhpsJ4TAa6+9hsmTJ2Pbtm3o1KkTVq1aVev6WRyoUTp8+DACAgLqfTtTp07Fli1b9C4LCgoy2ZDc9ODTN7jf1q1bERwcjKCgIMybNw9lZWV3PeBko/gjOCIlSkpKMnUEeoDd+Yd9586dw6ZNm7Bx40ZYWVnh/fffx5o1a9CuXbu7GnCSxYEavWPHjmH27NkIDQ3Fjz/+iCeeeAIXLlxA06ZNMWXKFKxbtw4XLlyAr68v5s2bZ3BdOTk5iIqKQm5uLp588knk5+dLy7p06YJ+/fohLS0NCQkJGDZsGA4dOoTw8HBMmDABfn5+AID33nsPAPDGG2/gm2++wX/+8x9oNBrY2tpi/vz5eOqppxAVFYXmzZvj7NmzUKvVcHNzw7vvvotmzZrV3xtFinb48GFkZWVhxIgRAG4PX965c2c4OTnhyJEjWL9+PTw8PLBs2TLEx8fXPuikiZ4jQWRSv/76qxg0aJA4dOiQ8PHxEampqeLXX38VnTp1EqdPnxZCCDFp0iQxcuRIUVZWJvLz84W7u7v0EJmahIeHi6VLlwohhMjMzBSenp7i22+/FUII4erqKhITE6W2rq6uIj8/X2zevFlMmTJFCCFEZWWl6NWrl7hw4YI4fPiwGD16tCguLhZCCPHzzz8Lf39/IYQQkZGRUrby8nIxZMgQsXnz5vv5FtEDwtvbW1y6dEmsXbtW58FGRUVF4saNGyIlJUUEBgZK88+dOycGDBhQ63p5zYEaLbVajWnTpsHHxwcdO3YEADg5OaFz584AgLZt2+L555+HpaUlWrZsiWbNmtV6l0dKSgpCQkIA3B5I7/nnn9dZ/uyzz8peM3DgQPz222/Iy8vDwYMH0a5dO7Rr1w779+9HVlYWQkNDERQUhPfeew+FhYW4fv06AKB3796wtLREkyZN4OrqarLHYZIyPP/889i9ezfy8/MhhMCbb76JL7744q4HnORpJWq0zM3NsWrVKoSHh8Pf3x8AYGlpqdPGwqJuXUSlUkFUG67sztfb2NjIXtO0aVP4+flh+/btOHHiBIYPHw7g9rMTgoKC8MYbb0jTubm5ePTRRwFA5ylxd26XGp+OHTtixowZGD9+PDQaDTp16oQpU6bAysrqrgac5JEDNVp2dnbo3r07IiMjMWfOHJSWlt7zOnv37o2vv/4aAPDXX3/h8OHDRr1uxIgRSExMxH//+1/p2kOvXr3w/fffIzc3FwDwn//8B+PHj7/njPRw2bdvH5ycnAAAw4cPx/bt27Fjxw68//77sLKyAgB07doVmzdvxvfff481a9agVatWta6XRw7U6AUHB2Pnzp2Ij4+Hubn5Pa0rNjYWc+fOxYABA+Do6CidrqpNly5dYG5uDn9/f6lD9+rVC5MnT8bEiROhUqnQvHlzrFixQpHPKKCHD4fsJiIiGR45ENVBRkYGIiIi9C5zcXHBsmXLGjYQUT3hkQMREcnwgjQREcmwOBARkQyLAxERybA4EBGRDIsDERHJ/B8FQ/2acnchpQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYcAAAEmCAYAAACJXlw1AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8QVMy6AAAACXBIWXMAAAsTAAALEwEAmpwYAAAs9ElEQVR4nO3deVQUV74H8G8DgqISJIJklDGgRCLRSDSOC4pGRRRaBDEiRlRGo4mij7gjLgliCMM8EwbNNp4XxxiXqLhFcR2XiMblGZe4JsiisiliC0gD3ff94aPHtppNaRq6v59zco5Vfbvqd/sSvtyq6iqZEEKAiIjoKWaGLoCIiBoehgMREUkwHIiISILhQEREEgwHIiKSYDgQEZEEw4Fq7Pbt23j99dfh7+8Pf39/yOVyjB49GufOndPL/jp16oT8/Pwq2xw5cgRffPHFC+1nwYIFWLNmTaX7v3TpEmbOnFnlNi5evIglS5a8UB0NxZo1a7BgwQJDl0EGZmHoAqhxadq0KXbs2KFZ3rNnDxYuXIj9+/cbpJ5Lly7h4cOHet1Hly5dkJCQUGWb33//HTk5OXqtg6g+MRzohRQUFMDe3l6zvGnTJqxbtw5mZmZo3bo1Fi9ejPbt22PSpElwd3fHvHnzkJKSggULFmDbtm2Ij4+HlZUVrl27hvv376Nv376IiopCkyZNtPazatUq/PTTTzA3N4ezszMWL16Mu3fvYuPGjVCpVGjZsiUiIiK03vPHH38gJiYGBQUFUKlUGD9+PIKCgmrdx19++QXR0dHYvXs3zp49i9jYWKjVagDA1KlT0bVrVyQkJODRo0dYuHAhPv30U52fg7OzM/Lz87Fw4UJkZGTA1tYW9vb2cHV1RXh4ON544w0MGjQI165dQ3x8PK5fv45NmzahrKwMDx8+xJQpUxASEoJt27Zh//79UKvVuHv3Ltq0aYN3330X33//PdLS0jBp0iSEhYWhuLgYy5YtQ3p6OgoKCtC8eXPEx8fDxcVFq39lZWVYvnw5UlJS8PLLL+Pll19Gy5YtAQCPHj1CTEwMbty4gbKyMvTu3Rvz5s2DhYX2r45bt27hk08+QVFREfLy8uDm5obPP/8cVlZWOHr0KOLj42FmZobXX38dKSkp+OGHH9CuXTv8+OOP2LBhA9RqNWxtbbF48WJ06NCh1mNEeiCIaigzM1O4ubmJESNGiBEjRogBAwYId3d3ceTIESGEECkpKWLw4MHi/v37Qgghtm7dKoYNGybUarXIyckRffr0EQcOHBD9+vUTp0+fFkIIMX/+fDFy5EhRWFgolEqlGDdunFi3bp0QQojXXntN3L9/X2zZskWMGTNGFBUVCSGESEhIEGFhYZp/f/zxx5Jay8rKxPDhw8Xly5eFEEIoFAoxbNgwcf78eUnb+fPnC09PT02/Kv6r2P+pU6eEr6+vEEKI0NBQsXv3biGEEFevXhXLli3T9PX999+v9nOIiIgQcXFxQgghcnJyRN++fUVCQoKmv0lJSUIIIQoLC8W7774r8vPzhRBCnD9/XnTr1k2zve7du4u7d+8KlUolhg8fLsLDw4VKpRJXr14VXbp0ESqVSuzdu1dER0dr+rl48WLxySefSPr/3XffidDQUKFUKkVRUZEICAgQ8+fPF0IIsWDBAvGvf/1LCCFEeXm5mDNnjvjmm28k24iNjRXbt28XQghRWloq/Pz8RHJyssjPzxc9e/YUV69eFUIIsW3bNvHaa6+JzMxM8csvv4iQkBBRXFwshBDi+PHjwsfHR7JtMgzOHKhWnj2slJKSgunTp2Pnzp04fvw4hg8fDjs7OwBAYGAgYmJicPv2bTg5OSE6OhoffvghwsPD8fbbb2u2ERAQgObNmwMA/P39cejQIbz33nua148dO4bAwEBYW1sDAEJDQ/HVV1+htLS00jrT0tKQkZGByMhIzbqSkhJcuXIF3bp1k7SfOHEi/vrXv2qt69Spk6TdsGHD8Mknn+Dw4cPo06cPPvroI0mbqj6Ho0ePIikpCQDg4OAAHx8frff26NEDANC8eXN89dVXOHr0KNLS0nDt2jUUFxdr2nXp0gWvvPIKAKBdu3bw9PSEmZkZnJycoFQq8fjxY/j4+MDJyQnr1q1Deno6Tp8+DQ8PD0m9J0+ehJ+fHywtLWFpaQm5XI7r168DeHJO59KlS9iyZYvmM9Rl7ty5OHHiBL799lukpaUhNzcXxcXFOHv2LDp06AA3NzcAT8Z6+fLlmm2np6cjODhYsx2FQoGCggLY2trq3A/VH4YDvZA+ffrgz3/+My5duqQ51PI0IQTKy8sBPDku37p1a1y8eFGrjbm5uVZ7MzPt6yTUajVkMpnWcsU2K1NxqOnpILt3757mcMnzCg4OxsCBA3HixAkcP34ciYmJSE5OltT7rIrPwcLCAuKp25k929eKAMzOzsaYMWPw7rvvonv37vDx8cG///1vTTtLS0ut9z17mAcAfvjhB2zevBnjxo2DXC6Hra0tbt++XW0fnx4PtVqNL774QnOoR6FQaI1FhY8++ggqlQrDhg3DgAEDkJWVBSEEzM3Ntfr7dJ/VajX8/f0xd+5czXJubi5eeumlamsk/ePVSvRCbt26hTt37uD1119Hv379sGfPHs0VRlu3boWtrS3at2+Pixcv4l//+he2bt2KR48eYe3atZpt7N27F6WlpVAqlUhKSsLAgQO19tGvXz9s3bpV85fzunXr8Pbbb8PS0hLm5uY6g8LZ2VlrlpOVlQU/Pz9cvnz5hfobHByMq1evIjAwENHR0VAoFMjLy9Oqo6rPwcvLS/NX+IMHD3Dw4EGdv2wvX74MOzs7fPjhh/D09NQEg0qlqnGtP//8MwICAjB69Gg4Ozvj8OHDOt/fr18/bN++HUqlEkqlEnv27NG85unpie+++w5CCJSWluKDDz7A999/r3Nf06dPx/DhwwEAFy5cgEqlwltvvaWZ+QDAvn37NAHj6emJn376Cbm5uQCADRs2YMKECTXuH+kXZw5UKyUlJfD399csq9VqfPLJJ3B2doazszMmTpyICRMmQK1Ww87ODl9//TWKi4vx0UcfISoqCm3atEFsbCxGjx6tObTUtGlThISEQKFQYOjQoRg1apTWPoOCgpCVlYXRo0dDrVajffv2iI+PBwD06tULc+bMQXR0NBYvXqx5j6WlJVavXo2YmBj885//RHl5OWbNmoXu3bu/UP/nzJmDFStW4PPPP4dMJsOMGTPQrl07qFQqrFq1CjNmzEBiYqLOz8HMzAwLFy5EVFSU5i/5P/3pT2jatKlkP3379sWWLVvg4+MDmUyGnj17ws7ODunp6TWuNSwsDEuWLNGEUbdu3XDjxg1Ju+DgYGRkZMDPz08TYhUWLVqEmJgYyOVylJWVoU+fPpg8ebJkGxEREZg+fTqsra3RokULvP3225qT7v/93/+N+fPnw8zMDG+88QYsLCzQrFkzeHp6YsqUKQgLC4NMJkOLFi2QmJioMyyp/snEs3M+onq0YMECuLq6So73G6v169ejc+fO8PDwQGlpKUJCQhAeHg4vLy9Dl6YXhYWFWL16NcLDw9GsWTP89ttvmDp1Ko4fP84QaOA4cyCqRx07dkR0dDTUajXKysrg4+NjtMEAAC1atECTJk0QFBQECwsLWFhYaGZd1LBx5kBERBI8IU1ERBIMByIikmA4EBGRBE9Ik8l59lvCJSUlCAkJ0boU9mkFBQVYtGgRTpw4gVatWuGjjz6CXC4H8OROtR9//DF+/fVXWFpaYujQoYiMjNT5pbT6MmfOHJw6dQrFxcWwt7fH5MmTMXr0aJ1tq6u/tp8VGRED3baDqEEoKioS3bp109zrSZeIiAgxa9YsUVhYKM6cOSPeeustcePGDSGEEJMnTxbz588XJSUlIjc3V/j5+Ym1a9fWV/k63bhxQyiVSiGEEL///rvo06ePuHTpks62tam/Jp8VGQ8eVqIG68svv8TSpUs1yw8fPoS7uzuUSmWd7WPfvn2ws7PT3NPoWcXFxdi/fz9mzZqF5s2bo0ePHnjnnXc037y+ffs2hg0bBisrK9jb28PT0xO///57jfZdVlaGlStX4p133oG7uzs6deqETp06YcSIES/UJ1dXV83tNWQyGWQyGTIyMnS2rU391X1WZFx4WIkarBs3buAvf/mLZvnq1atwdnaGlZWVVrupU6dW+sCh7t274+uvv650H0lJSRg5cmSl192npaXBzMwMzs7OmnVubm44c+YMgCc3Afzpp5/Qs2dPKBQKHD9+HLNmzapR/z7//HOcOXMG69evx0svvYQPP/wQLVq0wPz581+4f8uWLUNSUhJKSkrQuXPnSr9LUZv6q/usyLgwHKjBunHjBiZOnKhZvnbtms47pVb1y78qd+/exZkzZxATE1Npm+LiYsnN+lq2bImioiIAQM+ePfHjjz+ie/fuUKlUCAgIwODBg6vdd2FhIdatW4edO3dq7q7q7e2NvXv3wsnJSavt8/Rv2bJlWLx4Mc6fP4/Tp09LbtRXoab11+SzIuPCw0rUIJWWliIjIwOvvfaaZt21a9fw+uuv19k+tm/fju7du0t+GT/N2toahYWFWusKCwvRvHlzqNVq/PWvf8WQIUPw66+/4tSpU3j48CH+9re/Vbvvs2fPwsnJCa+++qpmnUKhQOvWrZ+7P88yNzdHjx49kJ2djQ0bNkher039NfmsyLgwHKhB+uOPP9CmTRs0a9YMwJNbXp8+fVrzXICnTZ48GR4eHjr/03WTuAo7duzAyJEjq6zj1VdfhUqlQlpammbdtWvX0LFjRxQUFCArKwvvvfceLC0t0apVK4waNQrHjh2rtn/5+fmwsbHRLAshcODAAQwYMKDO+ldBpVLpPOdQm/pr8lmRceFhJWqQrl+/jvv37yMjIwMODg748ssvcefOHbRt21bS9p///Gett/+///u/yMnJkTxs51nW1tYYMmQIEhISsHz5cly9ehWHDh3Cxo0bYWdnh3bt2mHDhg2ax3ImJSXpPPT1LFdXV1y5ckVzHqXibqQVt7x+3v7dv38fp06dwoABA9C0aVOkpKTgp59+0tzF9mk1rb+mnxUZF84cqEG6ceMG+vXrhylTpmDIkCFo3bo1nJyc8NVXX9XJ9rdv344hQ4agRYsWktcmT56stZ+lS5eipKQEffr0wezZs7Fs2TK4uroCABITE3H8+HH07t0bQ4YMgbm5udbT56ZMmYJDhw5J9tGlSxdMmzYNU6ZMweDBg3Hv3j188803kmdn15ZMJsOGDRvg5eWFt99+G3FxcYiMjNQ6j/B0/6qrv7rPiowXb7xHDVLFF7eGDh1q6FJeyObNm+Ho6Ij+/fsbuhSiWuHMgRqkGzduaB5N2ZiZm5ujd+/ehi6DqNY4c6AG5+HDh+jbty/Onz//wodZiOj5MByIiEiCh5WIiEiC4UBERBIMByIikmA4EBGRBMOBiIgkGA5ERCTBcCAiIgmGAxERSTAciIhIguFAREQSDAciIpLgw34aiG+//Rapqal62/6DBw8AAK1atdLbPmrDxcUFU6ZMMXQZRFQJhkMDkZqaistXrsO8qa1etq8qKQAAZD8o1cv2a6OiFiJquBgODYh5U1tYtx+kl20Xpz95Gpm+tl8bFbUQUcPFcw5ERCTBcCAiIgmGAxERSTAciIhIguFAREQSDAciIpJgOBARkQTDgYiIJBgOREQkwXAgIiIJhgMREUkwHIiISILhQEREEgwHIiKSYDgQEZEEw4GIiCQYDkREJGHS4XD48GEcPnzY0GUQPRf+/JI+mfRjQg8cOAAAeOeddwxcCVHt8eeX9MmkZw5ERKQbw4GIiCQYDkREJMFwICIiCYYDERFJMByIiEiC4UBERBIMByIikmA4EBGRBMOBiIgkGA5ERCTBcCAiIgmGAxERSTAciIhIguFAZCLkcrnmv+qMHTsWcrkcISEhNdr2rFmzIJfLERERUaP2e/bsgVwuR3JycrVto6OjIZfLERMTU6Ntf/bZZ5DL5fjb3/5W57UAwOeffw65XI6EhIRq26ampmLMmDG4detWjbadn5+PBQsW4MGDB3ppXxsMByKSKCwsBAA8evSoRu1TU1MBAL///nuN2n/11VcAgNWrV1fb9vTp0wCAU6dO1WjbP//8MwDg2LFjdV4LABw6dAjAf56nUZX4+HgUFxcjPj6+RtveuHEjrly5go0bN+qlfW0wHIhMwLOzhapmD2PHjtVarm72MGvWLK3l6mYPe/bsgRACACCEqPIv9ujoaK3l6mYPn332mdZydbOH2tQCPJk1PK2q2UNqaioyMzMBABkZGdXOHvLz83Ho0CEIIXDw4MFqZwO1bV9bJv0kuIKCAuTn52PhwoWGLgWpqalQl5sbuox6oS4vQWpqaoP43Buz1NRU2NnZ1fl2K2YNFaqbPVTMGipUN3uo+Eu9wurVq+Hj46OzbcWsoUJ1s4eKWUOFY8eOYe7cuXVSC/CfWUOFAwcOYObMmTrbPjtbiI+Px6pVqyrd9saNG6FWqwEAarUaGzduxAcffFBn7WuLMwciqlcVf6lXtlyf9FlLxayhQkZGRpXtjxw5gvLycgBAeXk5/v3vf9dp+9oy6ZmDra0tbG1t8emnnxq6FCxcuBBXU3MMXUa9MLNoCheXNg3ic2/MGuvMSyaTaf0SlslkRlmLk5OTVkD8+c9/rrL9gAEDcODAAZSXl8PCwgIDBw6s0/a1xZkDEWlp0aKF1nLLli2rbO/i4qK13LFjxyrbT5s2TWv5ww8/rLRtz549tZZ79epV5bY9PT21lvv3719ntQDAoEGDtJaHDBlSads5c+ZUufys4OBgmJk9+ZVsZmaG4ODgOm1fWwwHIhOwa9euKpeftmHDBq3lH374ocptf/HFF1rLK1eurLL98OHDNX+hy2SyKo/xL168WGt50aJFVW57/vz5WstVnW+obS0A8F//9V9ay5WdbwCehKaTkxOAJ7MGZ2fnKrdtZ2eHQYMGQSaTYfDgwWjVqlWdtq8thgMRSVTMHqqbNVSomD1UN2uoUPEXe3V/qQP/mT1UN2uoUDF7qG7W8Dy1AP+ZPVQ1a6gwZ84cWFtbVztrqBAcHIzOnTvXeBZQ2/a1IROGPBtkYBXHbBvCse+Kcw7W7QdV3/g5FKc/ucpCX9uvjeL0Q3id5xxeWEP6+SXjw5kDERFJMByIiEiC4UBERBIMByIikmA4EBGRBMOBiIgkGA5ERCTBcCAiIgmGAxERSTAciIhIguFAREQSDAciIpJgOBARkQTDgYiIJEz6MaE1uR87UUPFn1/SJ5MOh3feecfQJRA9N/78kj7xsBIREUkwHIiISILhQEREEgwHIiKSYDgQEZEEw4GIiCQYDkREJMFwICIiCYYDERFJMByIiEiC4UBERBIMByIikmA4EBGRBMOBiIgkGA5ERCTBcCAiIgmGAxERSZj0k+AaGlVJAYrTD+lt2wD0tv3aeFJLG0OXQURVYDg0EC4uLnrd/oMHlgCAVq1a6XU/NdNG7/0lohcjE0IIQxdBREQNC885EBGRBMOBiIgkGA5ERCTBcCAiIgmGAxERSTAciIhIguFAREQSjf5LcOXl5cjOzjZ0GUREjZKjoyMsLKRR0OjDITs7G4MGDTJ0GUREjdKhQ4fQrl07yfpG/w3pZ2cO2dnZGDduHNavXw9HR0cDVqZfptJPwHT6air9BEynr42hn0Y7c7CwsNCZeo6OjjrXGxtT6SdgOn01lX4CptPXxthPnpAmIiIJhgMREUkwHIiISMLowsHGxgYzZsyAjY2NoUvRK1PpJ2A6fTWVfgKm09fG3M9Gf7USERHVPaObORAR0YtjOBARkYRRhcOuXbswfPhweHt7Y/369YYuR2/Gjx8PX19f+Pv7w9/fHxcuXDB0SXWusLAQfn5+uH37NgAgJSUFcrkc3t7eWLlypYGrqzvP9nPhwoXw9vbWjO2BAwcMXGHdSExMhK+vL3x9fREXFwfAOMdUVz8b7ZgKI5GdnS0GDhwoHjx4IIqKioRcLhc3b940dFl1Tq1WC09PT1FWVmboUvTm119/FX5+fsLd3V1kZmaKx48fCy8vL5GRkSHKyspEWFiYOHLkiKHLfGHP9lMIIfz8/EROTo6BK6tbJ06cEGPGjBFKpVKUlpaK0NBQsWvXLqMbU1393L9/f6MdU6OZOaSkpKBXr16wtbWFtbU1hg4diuTkZEOXVedSU1MBAGFhYRgxYgS+//57A1dU9zZv3oylS5fCwcEBAHDx4kW0b98eTk5OsLCwgFwuN4qxfbafjx8/xt27dxEZGQm5XI6EhASo1WoDV/ni7O3tsWDBAlhaWqJJkybo0KED0tLSjG5MdfXz7t27jXZMjSYccnNzYW9vr1l2cHBATk6OASvSD4VCgd69e2PVqlX47rvvsHHjRpw4ccLQZdWpmJgY9OjRQ7NsrGP7bD/v3buHXr16YcWKFdi8eTPOnj2LLVu2GLDCuuHq6opu3boBANLS0rB3717IZDKjG1Nd/ezXr1+jHVOjCQe1Wg2ZTKZZFkJoLRsLDw8PxMXFoWXLlrCzs0NQUBCOHj1q6LL0ylTG1snJCatWrYKDgwOaNWuG8ePHG9XY3rx5E2FhYZg3bx6cnJyMdkyf7qeLi0ujHVOjCQdHR0fk5eVplvPy8jTTdWNy9uxZnDx5UrMshNB5R0VjYipje/36dezbt0+zbExje+7cOUycOBGzZ89GQECA0Y7ps/1szGNqNOHQp08fnDx5Evn5+Xj8+DH279+P/v37G7qsOvfo0SPExcVBqVSisLAQSUlJGDJkiKHL0qs333wTt27dQnp6OlQqFXbv3m2UYyuEwIoVK/Dw4UOUlZVh06ZNRjG2WVlZmD59OuLj4+Hr6wvAOMdUVz8b85g2jgirgTZt2iAiIgKhoaEoKytDUFAQunbtauiy6tzAgQNx4cIFjBw5Emq1GiEhIfDw8DB0WXplZWWF2NhYhIeHQ6lUwsvLCz4+PoYuq865ubnh/fffx9ixY1FeXg5vb2/4+fkZuqwXtmbNGiiVSsTGxmrWBQcHG92YVtbPxjqmjf72GRUP+6nsgRVERFR7jf6wUsVjQvkcaSKiutPow4GIiOoew4GIiCQYDkREJMFwICIiCV7e0wAoikrxuKRM52vNmjaBTXPLeq6IiEwdw6EBeFxShi82/arztVljujEciKje8bASERFJMByIiEiC4UBERBIMByIikmA4EBGRBMOBiIgkGA5ERCTBcCAiIgmGAxERSTAciIhIguFAREQSeg2Hw4cPIzAwEMOGDcPy5csBACkpKZDL5fD29sbKlSs1ba9evYrAwEAMHToUixYtQnl5uT5LIyKiKugtHDIzM7F06VKsXr0aO3fuxJUrV3D06FFERkZi9erV2LNnDy5fvoyjR48CAObOnYslS5Zg3759EEJg8+bN+iqNiIiqobdwOHDgAIYPHw5HR0c0adIEK1euRLNmzdC+fXs4OTnBwsICcrkcycnJuHPnDkpKStCtWzcAQGBgIJKTk/VVGhERVUNvt+xOT09HkyZNMG3aNGRlZWHAgAFwdXWFvb29po2DgwNycnKQm5urtd7e3h45OTmSbSoUCigUCq112dnZ+uoCEZHJ0ls4qFQqnD17FuvWrYO1tTU++OADNG3aFDKZTNNGCAGZTAa1Wq1z/bPWrl2LxMREfZVMRET/T2/h0Lp1a/Tu3Rt2dnYAgMGDByM5ORnm5uaaNnl5eXBwcICjoyPy8vI06+/duwcHBwfJNidMmICAgACtddnZ2Rg3bpyeetGwVfYEOT49johelN7CYeDAgZg/fz4UCgWaN2+O48ePw8fHB9988w3S09PRrl077N69G6NGjULbtm1hZWWFc+fOoXv37tixYwf69+8v2aaNjQ1sbGz0VXKjU9kT5Pj0OCJ6UXoLhzfffBOTJ09GSEgIysrK0LdvX4wdOxYuLi4IDw+HUqmEl5cXfHx8AADx8fGIiopCYWEh3N3dERoaqq/SiIioGnp9hnRQUBCCgoK01vXu3Rs7d+6UtHVzc8OWLVv0WQ4REdUQvyFNREQSDAciIpLQ62ElU1TZFURA/V1FZGYmQ879IoPWQESNG8OhjlV2BRFQf1cRKctUWL3lokFrIKLGjYeViIhIguFAREQSDAciIpJgOBARkQTDgYiIJGoUDpGRkZJ1M2fOrPNiiIioYajyUtalS5ciJycH586dQ35+vmZ9eXk5MjMz9V4cNXy8MyyRcaoyHIKCgnDz5k1cv34dQ4cO1aw3NzfXPLWNTBvvDEtknKoMhy5duqBLly7o06cPHB0d66smIiIysBp9QzorKwtz587Fw4cPIYTQrN+1a5feCiMiIsOpUTgsWbIEgYGB6Ny5s87HdxIRkXGpUThYWFhg0qRJ+q6FDKgh3DCQiBqOGoWDq6srrl+/jk6dOum7HjKQ+rxhIIOIqOGrUThkZmZi1KhR+NOf/gQrKyvN+pqcc/jss8/w4MEDxMbGIiUlBZ9++imUSiWGDRuGiIgIAMDVq1exaNEiFBUVoUePHvj4449hYcEbxhqrqoIoYqwHcnhpLJHB1eg3cMUv8do6efIkkpKSMGDAAJSUlCAyMhLr1q3DK6+8gqlTp+Lo0aPw8vLC3LlzsXz5cnTr1g2RkZHYvHkzQkJCnmuf1LhVdrtxXhpLVL9qFA6vvfZarTdcUFCAlStXYtq0abh27RouXryI9u3bw8nJCQAgl8uRnJyMjh07oqSkRPO9icDAQCQkJOgMB4VCAYVCobUuOzu71rUREVHVahQOvXr1gkwmgxBCc7WSvb09jh07Vul7lixZgoiICGRlZQEAcnNzYW9vr3ndwcEBOTk5kvX29vbIycnRuc21a9ciMTGxJiUTEdELqFE4XLt2TfPv0tJS7N69G7du3aq0/Y8//ohXXnkFvXv3xrZt2wAAarVa6zLYiqCpbL0uEyZMQEBAgNa67OxsjBs3ribdICKiGqr1WV9LS0sEBgYiMDAQs2fP1tlmz549yMvLg7+/Px4+fIji4mLcuXMH5ubmmjZ5eXlwcHCAo6Mj8vLyNOvv3bsHBwcHndu1sbGBjY1NbUsmIqJaqlE4FBQUaP4thMDly5clx/6f9j//8z+af2/btg2nT5/Gxx9/DG9vb6Snp6Ndu3bYvXs3Ro0ahbZt28LKygrnzp1D9+7dsWPHDvTv3//5e1SHeMklEZmqWp9zAICXX34ZixYtqtWOrKysEBsbi/DwcCiVSnh5ecHHxwcAEB8fj6ioKBQWFsLd3R2hoaG17IZ+1Oe1/0REDUmtzznUVsUhKADo3bs3du7cKWnj5uaGLVu2PPc+iIiobtUoHNRqNdasWYNjx46hvLwcffv2xbRp0/hFNSIiI1WjJ8H9/e9/x6lTpzBhwgRMmjQJ58+fR1xcnL5rIyIiA6nRn/7Hjx/H1q1b0aRJEwDAgAEDMGLECJ2PDyUiosavRjMHIYQmGIAnl7M+vUxERMalRuHg5uaGFStWICMjA5mZmVixYsVz3VKDiIgahxqFw9KlS6FQKBAcHIzRo0fjwYMHWLx4sb5rIyIiA6kyHEpLSzF//nycPHlSc8vtrl27wtzcHC1atKivGomIqJ5VGQ4JCQkoLCzEW2+9pVkXHR0NhUKBf/zjH3ovjoiIDKPKcDhy5Aj+/ve/4+WXX9asa9OmDeLi4nDw4EG9F0dERIZRZTg0adIETZs2laxv0aIFLC156wgiImNVZTiYmZmhsLBQsr6wsBDl5eV6K4qIiAyrynDw8/NDVFQUiouLNeuKi4sRFRUFb29vvRdHRESGUWU4TJgwAS1btkTfvn3x7rvvIigoCH379oWNjQ2mT59eXzUSEVE9q/L2GWZmZoiOjsa0adPw22+/wczMDF27dq30YTxEhlDZczf4zA2i51ejeyu1bdsWbdu21XctRM+lsudu8JkbRM+vRt+QJiIi08JwICIiCb2GQ2JiInx9feHr66t5/kNKSgrkcjm8vb2xcuVKTdurV68iMDAQQ4cOxaJFi3ipLBGRAektHFJSUvDzzz8jKSkJ27dvx2+//Ybdu3cjMjISq1evxp49e3D58mUcPXoUADB37lwsWbIE+/btgxACmzdv1ldpRERUDb2Fg729PRYsWKB59kOHDh2QlpaG9u3bw8nJCRYWFpDL5UhOTsadO3dQUlKCbt26AXjy3Onk5GTJNhUKBW7fvq31X3Z2tr66QERksvT2EGhXV1fNv9PS0rB371689957sLe316x3cHBATk4OcnNztdbb29sjJydHss21a9ciMTFRXyUTEdH/01s4VLh58yamTp2KefPmwdzcHGlpaZrXhBCQyWRQq9WQyWSS9c+aMGECAgICtNZlZ2dj3LhxequfiMgU6TUczp07h5kzZyIyMhK+vr44ffo08vLyNK/n5eXBwcEBjo6OWuvv3bun84t2NjY2sLGx0WfJREQEPZ5zyMrKwvTp0xEfHw9fX18AwJtvvolbt24hPT0dKpUKu3fvRv/+/dG2bVtYWVnh3LlzAIAdO3agf//++iqNiIiqobeZw5o1a6BUKhEbG6tZFxwcjNjYWISHh0OpVMLLyws+Pj4AgPj4eERFRaGwsBDu7u4IDQ3VV2lERFQNvYVDVFQUoqKidL62c+dOyTo3Nzds2bJFX+UQEVEt6P2ENFFDVNnN+gDesI8IYDiQiarsZn0Ab9hHBPDeSkREpANnDkQ1xENRZEoYDkQ1xENRZEoYDkR6xNkGNVYMByI94myDGiuekCYiIgmGAxERSTAciIhIguFAREQSPCFNZCBmZjLk3C/S+RqvZCJDYzgQGYiyTIXVWy7qfI1XMpGhMRyIGhF+b4Lqi8mHQ1X/s6nUop6rIara835vorKfcwYKVcbkw6Gq/9k+DOpav8UQ6UllP+c8fEWV4dVKREQk0aBmDrt27cKXX36J8vJyTJgwAePGjTN0SUQmi4eiTFuDCYecnBysXLkS27Ztg6WlJYKDg/GXv/wFHTt2NHRpRCapskNREWM9kMOT4kavwYRDSkoKevXqBVtbWwDA0KFDkZycjBkzZmjaKBQKKBQKrffduXMHAJCdnf1c+71X8BhFins6X8vOulvpa1lZd1H2uFmttleX76nqffVVd1Xve573VFV7fdZQX2PbWMcpPSMT6/Ze0/meSX6d0dpW9/aeR+HjMpQoyyXrm1pZoEWzJrV6T3Xvq20Nz7u9hsbR0REWFtIokAkhGsQlOV9//TWKi4sREREBAPjxxx9x8eJFREdHa9r84x//QGJioqFKJCIyOocOHUK7du0k6xvMzEGtVkMmk2mWhRBaywAwYcIEBAQEaK0rLS1FZmYmXn31VZibmyM7Oxvjxo3D+vXr4ejoWC+1G4Kp9BMwnb6aSj8B0+lrY+hnZXU1mHBwdHTE2bNnNct5eXlwcHDQamNjYwMbGxvJe11cXHRuT1caGhtT6SdgOn01lX4CptPXxtjPBnMpa58+fXDy5Enk5+fj8ePH2L9/P/r372/osoiITFKDmTm0adMGERERCA0NRVlZGYKCgtC1K7+ERkRkCA0mHABALpdDLpcbugwiIpPXYA4r1RUbGxvMmDFD57kJY2Iq/QRMp6+m0k/AdPramPvZYC5lJSKihsPoZg5ERPTiGA5ERCRhVOGwa9cuDB8+HN7e3li/fr2hy9Gb8ePHw9fXF/7+/vD398eFCxcMXVKdKywshJ+fH27fvg3gye1V5HI5vL29sXLlSgNXV3ee7efChQvh7e2tGdsDBw4YuMK6kZiYCF9fX/j6+iIuLg6AcY6prn422jEVRiI7O1sMHDhQPHjwQBQVFQm5XC5u3rxp6LLqnFqtFp6enqKsrMzQpejNr7/+Kvz8/IS7u7vIzMwUjx8/Fl5eXiIjI0OUlZWJsLAwceTIEUOX+cKe7acQQvj5+YmcnBwDV1a3Tpw4IcaMGSOUSqUoLS0VoaGhYteuXUY3prr6uX///kY7pkYzc3j6xn3W1taaG/cZm9TUVABAWFgYRowYge+//97AFdW9zZs3Y+nSpZpvyF+8eBHt27eHk5MTLCwsIJfLjWJsn+3n48ePcffuXURGRkIulyMhIQFqtdrAVb44e3t7LFiwAJaWlmjSpAk6dOiAtLQ0oxtTXf28e/duox1TowmH3Nxc2Nvba5YdHByQk5NjwIr0Q6FQoHfv3li1ahW+++47bNy4ESdOnDB0WXUqJiYGPXr00Cwb69g+28979+6hV69eWLFiBTZv3oyzZ89iy5YtBqywbri6uqJbt24AgLS0NOzduxcymczoxlRXP/v169dox9RowqEmN+4zBh4eHoiLi0PLli1hZ2eHoKAgHD161NBl6ZWpjK2TkxNWrVoFBwcHNGvWDOPHjzeqsb158ybCwsIwb948ODk5Ge2YPt1PFxeXRjumRhMOjo6OyMvL0yzrunGfMTh79ixOnjypWRZC6LwXuzExlbG9fv069u3bp1k2prE9d+4cJk6ciNmzZyMgIMBox/TZfjbmMTWacDCVG/c9evQIcXFxUCqVKCwsRFJSEoYMGWLosvTqzTffxK1bt5Ceng6VSoXdu3cb5dgKIbBixQo8fPgQZWVl2LRpk1GMbVZWFqZPn474+Hj4+voCMM4x1dXPxjymjSPCasBUbtw3cOBAXLhwASNHjoRarUZISAg8PDwMXZZeWVlZITY2FuHh4VAqlfDy8oKPj4+hy6pzbm5ueP/99zF27FiUl5fD29sbfn5+hi7rha1ZswZKpRKxsbGadcHBwUY3ppX1s7GOKW+fQUREEkZzWImIiOoOw4GIiCQYDkREJMFwICIiCYYDERFJMByIiEiC4UBERBJG8yU4ovqmVquxYsUKXLhwAUVFRRBCYPny5XB2dsbChQuRkZEBW1tb2Nvbw9XVFeHh4fjjjz8QExODgoICqFQqjB8/HkFBQYbuCpEEw4HoOV24cAG5ubnYtGkTzMzM8M033+Dbb7+FtbU1OnbsiK+//hq5ubkIDAyEq6srysvLMXPmTMTFxcHd3R2PHj3CmDFj0LFjR83dPIkaCoYD0XPy8PDASy+9hI0bNyIzMxO//PILmjdvjjNnziApKQnAk1tRV9wWIi0tDRkZGYiMjNRso6SkBFeuXGE4UIPDcCB6TkeOHEFMTAwmTZqEQYMGwcXFBTt37oSFhQWeviuNmdmTU3sqlQotW7bEjh07NK/du3cPLVu2rPfaiarDE9JEz+nEiRMYOHAgQkJC8MYbb+DgwYNQqVTw8vLSPNDlwYMHOHjwIGQyGZydndG0aVNNOGRlZcHPzw+XL182ZDeIdOKN94ie0x9//IHZs2dDpVKhvLwcffv2xf79+7Fjxw5ERUVpTkgLITBgwABMnjwZ165d05yQLi8vR2hoKMaOHWvorhBJMByI6tj69evRuXNneHh4oLS0FCEhIQgPD4eXl5ehSyOqMZ5zIKpjHTt2RHR0NNRqNcrKyuDj48NgoEaHMwciIpLgCWkiIpJgOBARkQTDgYiIJBgOREQkwXAgIiIJhgMREUn8HxlfSTkc2IZrAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "c = ['selling_price','km_driven','mileage','engine', 'max_power', 'seats', 'age']\n",
    "\n",
    "for i in c:\n",
    "\n",
    "  sns.set(style=\"ticks\")\n",
    "\n",
    "  x = df1[i]\n",
    "  coluna = i\n",
    "  mu = round(x.mean(),2) # mean of distribution\n",
    "  sigma = round(x.std(),2)  # standard deviation of distribution\n",
    "\n",
    "  f, (ax_box, ax_hist) = plt.subplots(2)\n",
    "\n",
    "  sns.boxplot(x=x, ax=ax_box)\n",
    "  sns.histplot(x=x, ax=ax_hist)\n",
    "\n",
    "  ax_box.set(yticks=[])\n",
    "  sns.despine(ax=ax_hist)\n",
    "  sns.despine(ax=ax_box, left=True)\n",
    "  ax_box.set_title('Boxplot e Histograma de {}\\n $\\mu={}$, $\\sigma={}$'.format(coluna, mu,sigma))\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "1N0OOpfAPInk"
   },
   "source": [
    "Nota se que a variavel que devemos prever, selling price tem uma altissima variancia e quase todas as variaveis quantitativas possuem outliers"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "COvVJeHx7O7J"
   },
   "source": [
    "## Preparacao dos dados"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Ix5D0hVs7Sat"
   },
   "source": [
    "antes de fazer a feature selection vamos normalizar e separar o que deve ser predito das variaveis <br>\n",
    "duvida: deve ser normalizado antes da feature selection?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 204
    },
    "id": "ZgSy4H1A6bix",
    "outputId": "7fbd4c89-9af6-4192-b1fa-ccfb0408f4d3"
   },
   "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>selling_price</th>\n",
       "      <th>km_driven</th>\n",
       "      <th>fuel</th>\n",
       "      <th>seller_type</th>\n",
       "      <th>transmission</th>\n",
       "      <th>owner</th>\n",
       "      <th>mileage</th>\n",
       "      <th>engine</th>\n",
       "      <th>max_power</th>\n",
       "      <th>seats</th>\n",
       "      <th>brand</th>\n",
       "      <th>age</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>459999</td>\n",
       "      <td>87000</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>20.77</td>\n",
       "      <td>1248</td>\n",
       "      <td>88.76</td>\n",
       "      <td>7.0</td>\n",
       "      <td>20</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1100000</td>\n",
       "      <td>102000</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>19.62</td>\n",
       "      <td>1995</td>\n",
       "      <td>187.74</td>\n",
       "      <td>5.0</td>\n",
       "      <td>3</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>229999</td>\n",
       "      <td>212000</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>11.57</td>\n",
       "      <td>2179</td>\n",
       "      <td>138.10</td>\n",
       "      <td>7.0</td>\n",
       "      <td>26</td>\n",
       "      <td>12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>800000</td>\n",
       "      <td>125000</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>11.50</td>\n",
       "      <td>2982</td>\n",
       "      <td>171.00</td>\n",
       "      <td>7.0</td>\n",
       "      <td>27</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>180000</td>\n",
       "      <td>25000</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>19.70</td>\n",
       "      <td>796</td>\n",
       "      <td>46.30</td>\n",
       "      <td>5.0</td>\n",
       "      <td>20</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   selling_price  km_driven fuel seller_type transmission owner  mileage  \\\n",
       "1         459999      87000    1           1            1     0    20.77   \n",
       "2        1100000     102000    1           0            0     0    19.62   \n",
       "3         229999     212000    1           1            1     4    11.57   \n",
       "4         800000     125000    1           1            1     2    11.50   \n",
       "5         180000      25000    3           1            1     2    19.70   \n",
       "\n",
       "   engine  max_power  seats brand  age  \n",
       "1    1248      88.76    7.0    20    9  \n",
       "2    1995     187.74    5.0     3   11  \n",
       "3    2179     138.10    7.0    26   12  \n",
       "4    2982     171.00    7.0    27   11  \n",
       "5     796      46.30    5.0    20   11  "
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df1.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "id": "LQQKL0gmd18L"
   },
   "outputs": [],
   "source": [
    "colunas = df1.iloc[:,1:].columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "_SeReaNcd18L",
    "outputId": "f1f6102a-b648-4da9-b871-bfb84d6f7e56"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['km_driven', 'fuel', 'seller_type', 'transmission', 'owner', 'mileage',\n",
       "       'engine', 'max_power', 'seats', 'brand', 'age'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "colunas"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ext0cWqB8cgt"
   },
   "source": [
    "### Padronizacao"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "CZ4C5y6O9aA2"
   },
   "source": [
    "Vamos padronizar a base a ser predita tbm"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "IsvhHjXn-o5R"
   },
   "source": [
    "Transformar para formato numpy para nao termos erro na normalizacao"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {
    "id": "ACl7ENAn-n9A"
   },
   "outputs": [],
   "source": [
    "data = df1.to_numpy()\n",
    "nrow,ncol = df1.shape\n",
    "y = data[:,:1]\n",
    "X = data[:,1:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "id": "iJsZnUw78OMB"
   },
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "\n",
    "scaler_train = StandardScaler()\n",
    "#scaler_train = MinMaxScaler()\n",
    "X = scaler_train.fit_transform(X)\n",
    "\n",
    "#Padronizando os precos tbm\n",
    "#scaler_train = StandardScaler()\n",
    "#scaler_train = MinMaxScaler()\n",
    "#y = scaler_train.fit_transform(y)\n",
    "\n",
    "#Vamos padronizar o teste tbm\n",
    "scaler_train = StandardScaler()\n",
    "#scaler_train = MinMaxScaler()\n",
    "df_test1 = scaler_train.fit_transform(df_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "NLazgycV_lMt",
    "outputId": "a5ce6fec-1ce7-47f3-c70b-0e20bca0c086"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(5530, 1)\n",
      "(5530, 11)\n"
     ]
    }
   ],
   "source": [
    "print(y.shape)\n",
    "print(X.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "JYaEaiFVABkO"
   },
   "source": [
    "### Separacao treino e teste"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "id": "YXq9T9u6_nPr"
   },
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.2, random_state = 4)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "JjpYugqyXiqi"
   },
   "source": [
    "## Catboost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "id": "d7sVQJfTd18O"
   },
   "outputs": [],
   "source": [
    "from catboost import CatBoostRegressor\n",
    "from sklearn.model_selection import GridSearchCV"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "id": "-09m2Xk4VaE5"
   },
   "outputs": [],
   "source": [
    " parameters = {'depth'         : [6,8,10],\n",
    "                  'learning_rate' : [0.01, 0.05, 0.1],\n",
    "                  'iterations'    : [30, 50, 100]\n",
    "                 }\n",
    "model_CBR = CatBoostRegressor(logging_level='Silent')\n",
    "\n",
    "eval_set=[(X_train, y_train), (X_test, y_test)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "0YdBmrweVXEL",
    "outputId": "3972120d-5206-49d7-8b7e-357a8c2b5888"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Melhor modelo: <catboost.core.CatBoostRegressor object at 0x0000021C7A608F10>\n",
      "Melhor score: 0.9636405747742944\n",
      "Wall time: 4min\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "grid = GridSearchCV(estimator=model_CBR, param_grid = parameters, cv = 10, n_jobs=-1, scoring='r2')\n",
    "grid.fit(X_train, y_train, eval_set=eval_set, early_stopping_rounds=10)\n",
    "#grid.fit(X_train, y_train)\n",
    "\n",
    "\n",
    "\n",
    "print(\"Melhor modelo: {}\".format(grid.best_estimator_))\n",
    "print(\"Melhor score: {}\".format(grid.best_score_))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "IU8Un4NLdD1q",
    "outputId": "624e751a-7ca1-472a-fd94-52649df8585f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9525834925856915\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import r2_score\n",
    "y_predict = grid.predict(X_test)\n",
    "\n",
    "#rmse = np.sqrt(mean_squared_error(y_test,y_linear_pred))\n",
    "r2 = r2_score(y_test,y_predict)\n",
    "print(r2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "LTxJ8eIPd18Q"
   },
   "source": [
    "## Feature Selection"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "RAfusSJJd18Q"
   },
   "source": [
    "https://www.analyseup.com/learn-python-for-data-science/python-random-forest-feature-importance-plot.html"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "id": "DB-oGTwUd18R"
   },
   "outputs": [],
   "source": [
    "def plot_feature_importance(importance,names,model_type):\n",
    "    \n",
    "    #Create arrays from feature importance and feature names\n",
    "    feature_importance = np.array(importance)\n",
    "    feature_names = np.array(names)\n",
    "    \n",
    "    #Create a DataFrame using a Dictionary\n",
    "    data={'feature_names':feature_names,'feature_importance':feature_importance}\n",
    "    fi_df = pd.DataFrame(data)\n",
    "    \n",
    "    #Sort the DataFrame in order decreasing feature importance\n",
    "    fi_df.sort_values(by=['feature_importance'], ascending=False,inplace=True)\n",
    "    \n",
    "    #Define size of bar plot\n",
    "    plt.figure(figsize=(10,8))\n",
    "    #Plot Searborn bar chart\n",
    "    sns.barplot(x=fi_df['feature_importance'], y=fi_df['feature_names'])\n",
    "    #Add chart labels\n",
    "    plt.title(model_type + 'FEATURE IMPORTANCE')\n",
    "    plt.xlabel('FEATURE IMPORTANCE')\n",
    "    plt.ylabel('FEATURE NAMES')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 518
    },
    "id": "1MBOGDiYd18R",
    "outputId": "f9bd20d3-d7ee-4be8-aa9f-eddcb40308ff"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_feature_importance(grid.best_estimator_.get_feature_importance(),colunas,'CATBOOST ')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "18Bt_fdad18S"
   },
   "source": [
    "### Vamos tirar as colunas com menos importancia"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "id": "_pPo9ZGVd18S"
   },
   "outputs": [],
   "source": [
    "df1.head()\n",
    "c = ['owner', 'seats', 'seller_type']\n",
    "df2 = df1.drop(labels = c, axis = 1)\n",
    "\n",
    "df_test2 = df_test.drop(labels = c, axis = 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 419
    },
    "id": "-Vtj_hPed18T",
    "outputId": "09a8690d-fdd8-470f-eb60-a1db5b23939e"
   },
   "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>selling_price</th>\n",
       "      <th>km_driven</th>\n",
       "      <th>fuel</th>\n",
       "      <th>transmission</th>\n",
       "      <th>mileage</th>\n",
       "      <th>engine</th>\n",
       "      <th>max_power</th>\n",
       "      <th>brand</th>\n",
       "      <th>age</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>459999</td>\n",
       "      <td>87000</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>20.77</td>\n",
       "      <td>1248</td>\n",
       "      <td>88.76</td>\n",
       "      <td>20</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1100000</td>\n",
       "      <td>102000</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>19.62</td>\n",
       "      <td>1995</td>\n",
       "      <td>187.74</td>\n",
       "      <td>3</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>229999</td>\n",
       "      <td>212000</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>11.57</td>\n",
       "      <td>2179</td>\n",
       "      <td>138.10</td>\n",
       "      <td>26</td>\n",
       "      <td>12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>800000</td>\n",
       "      <td>125000</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>11.50</td>\n",
       "      <td>2982</td>\n",
       "      <td>171.00</td>\n",
       "      <td>27</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>180000</td>\n",
       "      <td>25000</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>19.70</td>\n",
       "      <td>796</td>\n",
       "      <td>46.30</td>\n",
       "      <td>20</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5684</th>\n",
       "      <td>550000</td>\n",
       "      <td>20000</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>18.90</td>\n",
       "      <td>1197</td>\n",
       "      <td>82.00</td>\n",
       "      <td>11</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5685</th>\n",
       "      <td>360000</td>\n",
       "      <td>81000</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>19.01</td>\n",
       "      <td>1461</td>\n",
       "      <td>108.45</td>\n",
       "      <td>24</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5686</th>\n",
       "      <td>310000</td>\n",
       "      <td>70000</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>19.30</td>\n",
       "      <td>1248</td>\n",
       "      <td>73.90</td>\n",
       "      <td>20</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5687</th>\n",
       "      <td>650000</td>\n",
       "      <td>57000</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>23.65</td>\n",
       "      <td>1248</td>\n",
       "      <td>88.50</td>\n",
       "      <td>20</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5688</th>\n",
       "      <td>420000</td>\n",
       "      <td>90000</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>24.40</td>\n",
       "      <td>1120</td>\n",
       "      <td>71.01</td>\n",
       "      <td>11</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5530 rows × 9 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      selling_price  km_driven fuel transmission  mileage  engine  max_power  \\\n",
       "1            459999      87000    1            1    20.77    1248      88.76   \n",
       "2           1100000     102000    1            0    19.62    1995     187.74   \n",
       "3            229999     212000    1            1    11.57    2179     138.10   \n",
       "4            800000     125000    1            1    11.50    2982     171.00   \n",
       "5            180000      25000    3            1    19.70     796      46.30   \n",
       "...             ...        ...  ...          ...      ...     ...        ...   \n",
       "5684         550000      20000    3            1    18.90    1197      82.00   \n",
       "5685         360000      81000    1            1    19.01    1461     108.45   \n",
       "5686         310000      70000    1            1    19.30    1248      73.90   \n",
       "5687         650000      57000    1            1    23.65    1248      88.50   \n",
       "5688         420000      90000    1            1    24.40    1120      71.01   \n",
       "\n",
       "     brand  age  \n",
       "1       20    9  \n",
       "2        3   11  \n",
       "3       26   12  \n",
       "4       27   11  \n",
       "5       20   11  \n",
       "...    ...  ...  \n",
       "5684    11    4  \n",
       "5685    24    8  \n",
       "5686    20   10  \n",
       "5687    20    6  \n",
       "5688    11    7  \n",
       "\n",
       "[5530 rows x 9 columns]"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "id": "BtCfx2SUd18T"
   },
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "\n",
    "data = df2.to_numpy()\n",
    "nrow,ncol = df2.shape\n",
    "y = data[:,:1]\n",
    "X = data[:,1:]\n",
    "\n",
    "scaler_train = StandardScaler()\n",
    "#scaler_train = MinMaxScaler()\n",
    "X = scaler_train.fit_transform(X)\n",
    "\n",
    "#Vamos padronizar o teste tbm\n",
    "scaler_train = StandardScaler()\n",
    "#scaler_train = MinMaxScaler()\n",
    "df_test2 = scaler_train.fit_transform(df_test2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "id": "SWexPDOSd18T"
   },
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.2, random_state = 4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "id": "C4brA94xd18U"
   },
   "outputs": [],
   "source": [
    "parameters = {'depth'         : [6,8,10],\n",
    "                  'learning_rate' : [0.01, 0.05, 0.1],\n",
    "                  'iterations'    : [30, 50, 100]\n",
    "                 }\n",
    "model_CBR = CatBoostRegressor()\n",
    "\n",
    "eval_set=[(X_train, y_train), (X_test, y_test)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "_vwzeXXid18U",
    "outputId": "1d1613d5-b01b-422e-8fdf-61f836fdf0e2"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0:\tlearn: 725099.3360888\ttest: 725099.3360888\ttest1: 725838.2268251\tbest: 725838.2268251 (0)\ttotal: 41.3ms\tremaining: 4.09s\n",
      "1:\tlearn: 667962.7502607\ttest: 667962.7502607\ttest1: 670005.4649577\tbest: 670005.4649577 (1)\ttotal: 47.4ms\tremaining: 2.32s\n",
      "2:\tlearn: 618026.2240742\ttest: 618026.2240742\ttest1: 621886.9502452\tbest: 621886.9502452 (2)\ttotal: 54.6ms\tremaining: 1.76s\n",
      "3:\tlearn: 570637.0627416\ttest: 570637.0627416\ttest1: 578186.2612178\tbest: 578186.2612178 (3)\ttotal: 60.7ms\tremaining: 1.46s\n",
      "4:\tlearn: 527755.3300552\ttest: 527755.3300552\ttest1: 536947.9038419\tbest: 536947.9038419 (4)\ttotal: 67.1ms\tremaining: 1.27s\n",
      "5:\tlearn: 490277.1990580\ttest: 490277.1990580\ttest1: 502595.3096877\tbest: 502595.3096877 (5)\ttotal: 73.5ms\tremaining: 1.15s\n",
      "6:\tlearn: 453374.2521619\ttest: 453374.2521619\ttest1: 467777.7143207\tbest: 467777.7143207 (6)\ttotal: 80.3ms\tremaining: 1.07s\n",
      "7:\tlearn: 420792.5779912\ttest: 420792.5779912\ttest1: 437861.6946392\tbest: 437861.6946392 (7)\ttotal: 90.2ms\tremaining: 1.04s\n",
      "8:\tlearn: 393043.0440251\ttest: 393043.0440251\ttest1: 411090.6830783\tbest: 411090.6830783 (8)\ttotal: 97.3ms\tremaining: 984ms\n",
      "9:\tlearn: 367221.8236830\ttest: 367221.8236830\ttest1: 386919.3392056\tbest: 386919.3392056 (9)\ttotal: 107ms\tremaining: 966ms\n",
      "10:\tlearn: 343181.0694443\ttest: 343181.0694443\ttest1: 364792.0635716\tbest: 364792.0635716 (10)\ttotal: 118ms\tremaining: 957ms\n",
      "11:\tlearn: 322896.6381826\ttest: 322896.6381826\ttest1: 345040.7307973\tbest: 345040.7307973 (11)\ttotal: 125ms\tremaining: 916ms\n",
      "12:\tlearn: 302845.9618204\ttest: 302845.9618204\ttest1: 326860.5565415\tbest: 326860.5565415 (12)\ttotal: 135ms\tremaining: 902ms\n",
      "13:\tlearn: 286641.4715800\ttest: 286641.4715800\ttest1: 312219.6312316\tbest: 312219.6312316 (13)\ttotal: 143ms\tremaining: 879ms\n",
      "14:\tlearn: 270952.0744893\ttest: 270952.0744893\ttest1: 297170.7323797\tbest: 297170.7323797 (14)\ttotal: 153ms\tremaining: 869ms\n",
      "15:\tlearn: 258475.3925918\ttest: 258475.3925918\ttest1: 286340.8212579\tbest: 286340.8212579 (15)\ttotal: 164ms\tremaining: 863ms\n",
      "16:\tlearn: 246371.0312132\ttest: 246371.0312132\ttest1: 275079.9896599\tbest: 275079.9896599 (16)\ttotal: 172ms\tremaining: 840ms\n",
      "17:\tlearn: 233887.0889817\ttest: 233887.0889817\ttest1: 264771.1551038\tbest: 264771.1551038 (17)\ttotal: 183ms\tremaining: 832ms\n",
      "18:\tlearn: 222216.5469460\ttest: 222216.5469460\ttest1: 255136.6229553\tbest: 255136.6229553 (18)\ttotal: 189ms\tremaining: 807ms\n",
      "19:\tlearn: 212494.4651008\ttest: 212494.4651008\ttest1: 246835.1105830\tbest: 246835.1105830 (19)\ttotal: 198ms\tremaining: 793ms\n",
      "20:\tlearn: 204382.3914991\ttest: 204382.3914991\ttest1: 240594.7171910\tbest: 240594.7171910 (20)\ttotal: 206ms\tremaining: 775ms\n",
      "21:\tlearn: 197076.2687288\ttest: 197076.2687288\ttest1: 234047.3517937\tbest: 234047.3517937 (21)\ttotal: 214ms\tremaining: 759ms\n",
      "22:\tlearn: 190717.9316182\ttest: 190717.9316182\ttest1: 229118.7773587\tbest: 229118.7773587 (22)\ttotal: 224ms\tremaining: 750ms\n",
      "23:\tlearn: 184265.7333922\ttest: 184265.7333922\ttest1: 224184.7455733\tbest: 224184.7455733 (23)\ttotal: 233ms\tremaining: 738ms\n",
      "24:\tlearn: 179223.7916875\ttest: 179223.7916875\ttest1: 219812.4763748\tbest: 219812.4763748 (24)\ttotal: 240ms\tremaining: 719ms\n",
      "25:\tlearn: 174527.9822117\ttest: 174527.9822117\ttest1: 214888.9196695\tbest: 214888.9196695 (25)\ttotal: 247ms\tremaining: 704ms\n",
      "26:\tlearn: 170404.2989815\ttest: 170404.2989815\ttest1: 211053.6780763\tbest: 211053.6780763 (26)\ttotal: 254ms\tremaining: 687ms\n",
      "27:\tlearn: 166863.7350093\ttest: 166863.7350093\ttest1: 208784.3367464\tbest: 208784.3367464 (27)\ttotal: 261ms\tremaining: 672ms\n",
      "28:\tlearn: 163260.1751759\ttest: 163260.1751759\ttest1: 205808.8396433\tbest: 205808.8396433 (28)\ttotal: 270ms\tremaining: 661ms\n",
      "29:\tlearn: 158926.3923587\ttest: 158926.3923587\ttest1: 202478.9809685\tbest: 202478.9809685 (29)\ttotal: 277ms\tremaining: 647ms\n",
      "30:\tlearn: 155785.3469715\ttest: 155785.3469715\ttest1: 200400.2700177\tbest: 200400.2700177 (30)\ttotal: 285ms\tremaining: 635ms\n",
      "31:\tlearn: 152749.5027848\ttest: 152749.5027848\ttest1: 198254.6641147\tbest: 198254.6641147 (31)\ttotal: 293ms\tremaining: 623ms\n",
      "32:\tlearn: 150483.6658493\ttest: 150483.6658493\ttest1: 196047.9345613\tbest: 196047.9345613 (32)\ttotal: 300ms\tremaining: 609ms\n",
      "33:\tlearn: 148229.8724341\ttest: 148229.8724341\ttest1: 194514.8600443\tbest: 194514.8600443 (33)\ttotal: 308ms\tremaining: 597ms\n",
      "34:\tlearn: 146027.7013355\ttest: 146027.7013355\ttest1: 192266.2622987\tbest: 192266.2622987 (34)\ttotal: 314ms\tremaining: 584ms\n",
      "35:\tlearn: 144245.4701191\ttest: 144245.4701191\ttest1: 191552.0168309\tbest: 191552.0168309 (35)\ttotal: 322ms\tremaining: 572ms\n",
      "36:\tlearn: 142413.2536117\ttest: 142413.2536117\ttest1: 190117.7821396\tbest: 190117.7821396 (36)\ttotal: 331ms\tremaining: 563ms\n",
      "37:\tlearn: 140688.3204281\ttest: 140688.3204281\ttest1: 188455.4328949\tbest: 188455.4328949 (37)\ttotal: 340ms\tremaining: 554ms\n",
      "38:\tlearn: 139066.2852420\ttest: 139066.2852420\ttest1: 187131.2301431\tbest: 187131.2301431 (38)\ttotal: 346ms\tremaining: 541ms\n",
      "39:\tlearn: 137603.2707736\ttest: 137603.2707736\ttest1: 185678.7582810\tbest: 185678.7582810 (39)\ttotal: 353ms\tremaining: 529ms\n",
      "40:\tlearn: 136068.0198434\ttest: 136068.0198434\ttest1: 184205.7321537\tbest: 184205.7321537 (40)\ttotal: 362ms\tremaining: 521ms\n",
      "41:\tlearn: 134378.2331538\ttest: 134378.2331538\ttest1: 183290.2839965\tbest: 183290.2839965 (41)\ttotal: 371ms\tremaining: 512ms\n",
      "42:\tlearn: 132782.4541325\ttest: 132782.4541325\ttest1: 182169.0617821\tbest: 182169.0617821 (42)\ttotal: 378ms\tremaining: 501ms\n",
      "43:\tlearn: 131790.3338328\ttest: 131790.3338328\ttest1: 181371.1566799\tbest: 181371.1566799 (43)\ttotal: 386ms\tremaining: 492ms\n",
      "44:\tlearn: 130880.2599549\ttest: 130880.2599549\ttest1: 180683.8795768\tbest: 180683.8795768 (44)\ttotal: 393ms\tremaining: 480ms\n",
      "45:\tlearn: 129808.2123748\ttest: 129808.2123748\ttest1: 180036.1907090\tbest: 180036.1907090 (45)\ttotal: 399ms\tremaining: 468ms\n",
      "46:\tlearn: 129303.5785381\ttest: 129303.5785381\ttest1: 179738.2600447\tbest: 179738.2600447 (46)\ttotal: 407ms\tremaining: 459ms\n",
      "47:\tlearn: 128048.3689670\ttest: 128048.3689670\ttest1: 178556.9712466\tbest: 178556.9712466 (47)\ttotal: 415ms\tremaining: 449ms\n",
      "48:\tlearn: 127037.1538879\ttest: 127037.1538879\ttest1: 178171.7057705\tbest: 178171.7057705 (48)\ttotal: 421ms\tremaining: 438ms\n",
      "49:\tlearn: 126176.0556530\ttest: 126176.0556530\ttest1: 177706.5702217\tbest: 177706.5702217 (49)\ttotal: 427ms\tremaining: 427ms\n",
      "50:\tlearn: 125570.3422714\ttest: 125570.3422714\ttest1: 177407.1423650\tbest: 177407.1423650 (50)\ttotal: 439ms\tremaining: 421ms\n",
      "51:\tlearn: 124650.7426271\ttest: 124650.7426271\ttest1: 176796.2265874\tbest: 176796.2265874 (51)\ttotal: 450ms\tremaining: 415ms\n",
      "52:\tlearn: 123882.3428112\ttest: 123882.3428112\ttest1: 176625.5441358\tbest: 176625.5441358 (52)\ttotal: 457ms\tremaining: 406ms\n",
      "53:\tlearn: 122924.4547779\ttest: 122924.4547779\ttest1: 175859.5014073\tbest: 175859.5014073 (53)\ttotal: 466ms\tremaining: 397ms\n",
      "54:\tlearn: 122129.0599356\ttest: 122129.0599356\ttest1: 175180.7398020\tbest: 175180.7398020 (54)\ttotal: 473ms\tremaining: 387ms\n",
      "55:\tlearn: 121511.6989019\ttest: 121511.6989019\ttest1: 174846.2723380\tbest: 174846.2723380 (55)\ttotal: 481ms\tremaining: 378ms\n",
      "56:\tlearn: 120589.4171994\ttest: 120589.4171994\ttest1: 174736.4785205\tbest: 174736.4785205 (56)\ttotal: 488ms\tremaining: 368ms\n",
      "57:\tlearn: 120008.4701474\ttest: 120008.4701474\ttest1: 174249.8404187\tbest: 174249.8404187 (57)\ttotal: 497ms\tremaining: 360ms\n",
      "58:\tlearn: 119430.6644494\ttest: 119430.6644494\ttest1: 174265.5090162\tbest: 174249.8404187 (57)\ttotal: 504ms\tremaining: 350ms\n",
      "59:\tlearn: 118630.3058181\ttest: 118630.3058181\ttest1: 173976.2184429\tbest: 173976.2184429 (59)\ttotal: 512ms\tremaining: 341ms\n",
      "60:\tlearn: 118182.1997115\ttest: 118182.1997115\ttest1: 173876.2073501\tbest: 173876.2073501 (60)\ttotal: 519ms\tremaining: 332ms\n",
      "61:\tlearn: 117570.8979635\ttest: 117570.8979635\ttest1: 173793.8082586\tbest: 173793.8082586 (61)\ttotal: 526ms\tremaining: 323ms\n",
      "62:\tlearn: 116938.2068202\ttest: 116938.2068202\ttest1: 173418.3256343\tbest: 173418.3256343 (62)\ttotal: 534ms\tremaining: 313ms\n",
      "63:\tlearn: 116403.4637366\ttest: 116403.4637366\ttest1: 173307.5139834\tbest: 173307.5139834 (63)\ttotal: 541ms\tremaining: 304ms\n",
      "64:\tlearn: 115943.1552822\ttest: 115943.1552822\ttest1: 172843.7260659\tbest: 172843.7260659 (64)\ttotal: 548ms\tremaining: 295ms\n",
      "65:\tlearn: 115501.0159642\ttest: 115501.0159642\ttest1: 172739.4278919\tbest: 172739.4278919 (65)\ttotal: 554ms\tremaining: 286ms\n",
      "66:\tlearn: 114896.8633806\ttest: 114896.8633806\ttest1: 172469.6259225\tbest: 172469.6259225 (66)\ttotal: 561ms\tremaining: 276ms\n",
      "67:\tlearn: 114371.1003208\ttest: 114371.1003208\ttest1: 172089.9094769\tbest: 172089.9094769 (67)\ttotal: 567ms\tremaining: 267ms\n",
      "68:\tlearn: 113769.1552686\ttest: 113769.1552686\ttest1: 171975.3847819\tbest: 171975.3847819 (68)\ttotal: 573ms\tremaining: 257ms\n",
      "69:\tlearn: 113367.6082125\ttest: 113367.6082125\ttest1: 171941.0547624\tbest: 171941.0547624 (69)\ttotal: 579ms\tremaining: 248ms\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "70:\tlearn: 112807.8812827\ttest: 112807.8812827\ttest1: 171750.1811411\tbest: 171750.1811411 (70)\ttotal: 587ms\tremaining: 240ms\n",
      "71:\tlearn: 112486.8281573\ttest: 112486.8281573\ttest1: 171731.0786620\tbest: 171731.0786620 (71)\ttotal: 594ms\tremaining: 231ms\n",
      "72:\tlearn: 111984.3409624\ttest: 111984.3409624\ttest1: 171211.1273708\tbest: 171211.1273708 (72)\ttotal: 600ms\tremaining: 222ms\n",
      "73:\tlearn: 111271.7262892\ttest: 111271.7262892\ttest1: 171116.4885800\tbest: 171116.4885800 (73)\ttotal: 607ms\tremaining: 213ms\n",
      "74:\tlearn: 110412.4742720\ttest: 110412.4742720\ttest1: 170661.2066491\tbest: 170661.2066491 (74)\ttotal: 613ms\tremaining: 204ms\n",
      "75:\tlearn: 109833.7743343\ttest: 109833.7743343\ttest1: 170522.0560090\tbest: 170522.0560090 (75)\ttotal: 621ms\tremaining: 196ms\n",
      "76:\tlearn: 109327.5292193\ttest: 109327.5292193\ttest1: 170162.1461219\tbest: 170162.1461219 (76)\ttotal: 629ms\tremaining: 188ms\n",
      "77:\tlearn: 108938.8797941\ttest: 108938.8797941\ttest1: 170065.9735423\tbest: 170065.9735423 (77)\ttotal: 637ms\tremaining: 180ms\n",
      "78:\tlearn: 108377.0521836\ttest: 108377.0521836\ttest1: 169758.9858351\tbest: 169758.9858351 (78)\ttotal: 643ms\tremaining: 171ms\n",
      "79:\tlearn: 107920.0212175\ttest: 107920.0212175\ttest1: 169823.8961036\tbest: 169758.9858351 (78)\ttotal: 649ms\tremaining: 162ms\n",
      "80:\tlearn: 107605.6619060\ttest: 107605.6619060\ttest1: 169663.5161577\tbest: 169663.5161577 (80)\ttotal: 657ms\tremaining: 154ms\n",
      "81:\tlearn: 107217.1526180\ttest: 107217.1526180\ttest1: 169521.3574243\tbest: 169521.3574243 (81)\ttotal: 665ms\tremaining: 146ms\n",
      "82:\tlearn: 106703.3143505\ttest: 106703.3143505\ttest1: 169183.0635164\tbest: 169183.0635164 (82)\ttotal: 672ms\tremaining: 138ms\n",
      "83:\tlearn: 106394.4399996\ttest: 106394.4399996\ttest1: 168881.9310923\tbest: 168881.9310923 (83)\ttotal: 678ms\tremaining: 129ms\n",
      "84:\tlearn: 105989.4218876\ttest: 105989.4218876\ttest1: 168980.7043083\tbest: 168881.9310923 (83)\ttotal: 686ms\tremaining: 121ms\n",
      "85:\tlearn: 105583.7281376\ttest: 105583.7281376\ttest1: 168676.7258927\tbest: 168676.7258927 (85)\ttotal: 692ms\tremaining: 113ms\n",
      "86:\tlearn: 104871.9827146\ttest: 104871.9827146\ttest1: 168297.8732113\tbest: 168297.8732113 (86)\ttotal: 698ms\tremaining: 104ms\n",
      "87:\tlearn: 104424.8228558\ttest: 104424.8228558\ttest1: 168045.4120648\tbest: 168045.4120648 (87)\ttotal: 705ms\tremaining: 96.1ms\n",
      "88:\tlearn: 103889.3089230\ttest: 103889.3089230\ttest1: 167956.6055735\tbest: 167956.6055735 (88)\ttotal: 711ms\tremaining: 87.9ms\n",
      "89:\tlearn: 103616.4776767\ttest: 103616.4776767\ttest1: 167977.5844983\tbest: 167956.6055735 (88)\ttotal: 719ms\tremaining: 79.8ms\n",
      "90:\tlearn: 103253.3979466\ttest: 103253.3979466\ttest1: 167773.6376465\tbest: 167773.6376465 (90)\ttotal: 725ms\tremaining: 71.7ms\n",
      "91:\tlearn: 102986.4402999\ttest: 102986.4402999\ttest1: 167711.6607700\tbest: 167711.6607700 (91)\ttotal: 733ms\tremaining: 63.7ms\n",
      "92:\tlearn: 102701.8595845\ttest: 102701.8595845\ttest1: 167547.6694351\tbest: 167547.6694351 (92)\ttotal: 740ms\tremaining: 55.7ms\n",
      "93:\tlearn: 102343.2845205\ttest: 102343.2845205\ttest1: 167279.3629962\tbest: 167279.3629962 (93)\ttotal: 748ms\tremaining: 47.8ms\n",
      "94:\tlearn: 101981.3584052\ttest: 101981.3584052\ttest1: 167225.8302052\tbest: 167225.8302052 (94)\ttotal: 755ms\tremaining: 39.7ms\n",
      "95:\tlearn: 101336.8314230\ttest: 101336.8314230\ttest1: 166898.4084543\tbest: 166898.4084543 (95)\ttotal: 763ms\tremaining: 31.8ms\n",
      "96:\tlearn: 100996.5880829\ttest: 100996.5880829\ttest1: 166986.0832879\tbest: 166898.4084543 (95)\ttotal: 771ms\tremaining: 23.8ms\n",
      "97:\tlearn: 100782.5086486\ttest: 100782.5086486\ttest1: 166796.3737366\tbest: 166796.3737366 (97)\ttotal: 778ms\tremaining: 15.9ms\n",
      "98:\tlearn: 100484.1626654\ttest: 100484.1626654\ttest1: 166742.1681268\tbest: 166742.1681268 (98)\ttotal: 785ms\tremaining: 7.93ms\n",
      "99:\tlearn: 100220.7995788\ttest: 100220.7995788\ttest1: 166791.4282489\tbest: 166742.1681268 (98)\ttotal: 793ms\tremaining: 0us\n",
      "\n",
      "bestTest = 166742.1681\n",
      "bestIteration = 98\n",
      "\n",
      "Shrink model to first 99 iterations.\n",
      "Melhor modelo: <catboost.core.CatBoostRegressor object at 0x0000021C7C5B27F0>\n",
      "Melhor score: 0.9645632608480236\n",
      "Wall time: 3min 52s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "grid = GridSearchCV(estimator=model_CBR, param_grid = parameters, cv = 10, n_jobs=-1, scoring='r2')\n",
    "grid.fit(X_train, y_train, eval_set=eval_set, early_stopping_rounds=10)\n",
    "#grid.fit(X_train, y_train)\n",
    "\n",
    "y_predict = grid.predict(X_test)\n",
    "\n",
    "#rmse = np.sqrt(mean_squared_error(y_test,y_linear_pred))\n",
    "#r2 = r2_score(y_test,y_linear_pred)\n",
    "\n",
    "print(\"Melhor modelo: {}\".format(grid.best_estimator_))\n",
    "print(\"Melhor score: {}\".format(grid.best_score_))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "aqDxFIpQ8LdQ"
   },
   "source": [
    "## Gradient Boosting"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "id": "YWE88gKW7Ysv"
   },
   "outputs": [],
   "source": [
    "from sklearn.ensemble import GradientBoostingRegressor\n",
    "from sklearn.model_selection import GridSearchCV"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "xipv09JM7LhF",
    "outputId": "ebdb542a-9b60-489c-8163-6a3188003372"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 10 folds for each of 1 candidates, totalling 10 fits\n",
      "Melhor modelo: GradientBoostingRegressor(learning_rate=0.02, max_depth=4, n_estimators=3000)\n",
      "Melhor score: 0.973730974769129\n",
      "Wall time: 1min 43s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "parameters = {'max_depth':[4], 'learning_rate':[0.02],\n",
    "             \"n_estimators\":[3000], \"loss\":[\"ls\"], \n",
    "              \"criterion\":[\"friedman_mse\"]}\n",
    "\n",
    "grb_model = GridSearchCV(GradientBoostingRegressor(), parameters,\n",
    "                    cv = 10, scoring = \"r2\", n_jobs = -1, verbose = 3,\n",
    "                    refit = True)\n",
    "\n",
    "grb_model.fit(X_train, y_train.ravel())\n",
    "y_pred_train = grb_model.predict(X_train)\n",
    "\n",
    "\n",
    "print(\"Melhor modelo: {}\".format(grb_model.best_estimator_))\n",
    "print(\"Melhor score: {}\".format(grb_model.best_score_))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "TJt_gxp4fgZ7"
   },
   "source": [
    "## Submetendo a predicao"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "id": "kVXOOB3TYOnc"
   },
   "outputs": [],
   "source": [
    "y_pred = grid.best_estimator_.predict(df_test2)\n",
    "y_pred = np.array(y_pred, dtype = int)\n",
    "prediction = pd.DataFrame()\n",
    "prediction['Id'] = Id\n",
    "prediction['selling_price'] = y_pred\n",
    "\n",
    "prediction.to_csv('catboost2.csv', index = False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 359
    },
    "id": "mG3IJCgwdpde",
    "outputId": "dc3e1aa3-8ac7-44de-a384-b993ccc3a210"
   },
   "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>selling_price</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>193067</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>556582</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>616575</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>1429865</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>554934</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>6</td>\n",
       "      <td>457523</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>7</td>\n",
       "      <td>652911</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>8</td>\n",
       "      <td>1964215</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>9</td>\n",
       "      <td>2390847</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>10</td>\n",
       "      <td>512478</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Id  selling_price\n",
       "0   1         193067\n",
       "1   2         556582\n",
       "2   3         616575\n",
       "3   4        1429865\n",
       "4   5         554934\n",
       "5   6         457523\n",
       "6   7         652911\n",
       "7   8        1964215\n",
       "8   9        2390847\n",
       "9  10         512478"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "prediction.head(10)"
   ]
  }
 ],
 "metadata": {
  "colab": {
   "name": "Car Price Final 2.ipynb",
   "provenance": []
  },
  "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": "3.8.8"
  },
  "latex_envs": {
   "LaTeX_envs_menu_present": true,
   "autoclose": false,
   "autocomplete": false,
   "bibliofile": "biblio.bib",
   "cite_by": "apalike",
   "current_citInitial": 1,
   "eqLabelWithNumbers": true,
   "eqNumInitial": 1,
   "hotkeys": {
    "equation": "Ctrl-E",
    "itemize": "Ctrl-I"
   },
   "labels_anchors": false,
   "latex_user_defs": false,
   "report_style_numbering": false,
   "user_envs_cfg": false
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}