Clube da Aposta.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Importação das bibliotecas necessárias\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "import warnings\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "import seaborn as sns\n",
    "\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.pipeline import make_pipeline, Pipeline\n",
    "from sklearn.model_selection import GridSearchCV, train_test_split\n",
    "from sklearn.metrics import confusion_matrix, classification_report, accuracy_score\n",
    "\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.ensemble import RandomForestClassifier, GradientBoostingClassifier\n",
    "\n",
    "from pickle import dump, load"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Ignorar futuros warnings. Vale ressaltar que eu já rodei sem a presença dessa linha e nenhum warning afeta os resultados\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Importando o conjunto de dados\n",
    "df = pd.read_json('teste.json')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "ordens_coluna = ['home_name', 'away_name', 'home_score', 'away_score', 'final_result',\n",
    "       'time', 'home_pos', 'away_pos', 'round', 'home_last5all_home',\n",
    "       'home_last5all_home_win', 'home_last5all_home_draw',\n",
    "       'home_last5all_home_lose', 'away_last5all_away',\n",
    "       'away_last5all_away_win', 'away_last5all_away_draw',\n",
    "       'away_last5all_away_lose', 'last5all_home_away_dif', 'fifa_home_ova',\n",
    "       'fifa_home_att', 'fifa_home_mid', 'fifa_home_def', 'fifa_away_ova',\n",
    "       'fifa_away_att', 'fifa_away_mid', 'fifa_away_def', 'elo_home_score',\n",
    "       'elo_away_score', 'tfm_value_home', 'tfm_value_away']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Alterando a ordem de colunas do df\n",
    "df = df[ordens_coluna]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h2>Análise Exploratória</h2>\n",
    "<h3>Hipótese</h3>\n",
    "1 - O elenco mais caro ganha mais os jogos (final_result)?<br/>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>home_name</th>\n",
       "      <th>away_name</th>\n",
       "      <th>home_score</th>\n",
       "      <th>away_score</th>\n",
       "      <th>final_result</th>\n",
       "      <th>time</th>\n",
       "      <th>home_pos</th>\n",
       "      <th>away_pos</th>\n",
       "      <th>round</th>\n",
       "      <th>home_last5all_home</th>\n",
       "      <th>...</th>\n",
       "      <th>fifa_home_mid</th>\n",
       "      <th>fifa_home_def</th>\n",
       "      <th>fifa_away_ova</th>\n",
       "      <th>fifa_away_att</th>\n",
       "      <th>fifa_away_mid</th>\n",
       "      <th>fifa_away_def</th>\n",
       "      <th>elo_home_score</th>\n",
       "      <th>elo_away_score</th>\n",
       "      <th>tfm_value_home</th>\n",
       "      <th>tfm_value_away</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>FC Porto</td>\n",
       "      <td>Nacional</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>H</td>\n",
       "      <td>1546895700</td>\n",
       "      <td>1</td>\n",
       "      <td>11</td>\n",
       "      <td>16</td>\n",
       "      <td>9</td>\n",
       "      <td>...</td>\n",
       "      <td>81</td>\n",
       "      <td>80</td>\n",
       "      <td>70</td>\n",
       "      <td>70</td>\n",
       "      <td>70</td>\n",
       "      <td>69</td>\n",
       "      <td>1813.906738</td>\n",
       "      <td>1392.986084</td>\n",
       "      <td>2.779800e+08</td>\n",
       "      <td>9450000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Tenerife</td>\n",
       "      <td>Elche</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>H</td>\n",
       "      <td>1546632000</td>\n",
       "      <td>18</td>\n",
       "      <td>15</td>\n",
       "      <td>20</td>\n",
       "      <td>2</td>\n",
       "      <td>...</td>\n",
       "      <td>71</td>\n",
       "      <td>69</td>\n",
       "      <td>67</td>\n",
       "      <td>69</td>\n",
       "      <td>67</td>\n",
       "      <td>67</td>\n",
       "      <td>1513.179932</td>\n",
       "      <td>1475.512939</td>\n",
       "      <td>1.090000e+07</td>\n",
       "      <td>9600000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>Man City</td>\n",
       "      <td>Everton</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>H</td>\n",
       "      <td>1544877000</td>\n",
       "      <td>2</td>\n",
       "      <td>7</td>\n",
       "      <td>17</td>\n",
       "      <td>16</td>\n",
       "      <td>...</td>\n",
       "      <td>88</td>\n",
       "      <td>83</td>\n",
       "      <td>79</td>\n",
       "      <td>79</td>\n",
       "      <td>80</td>\n",
       "      <td>79</td>\n",
       "      <td>1986.672974</td>\n",
       "      <td>1699.472778</td>\n",
       "      <td>1.090000e+09</td>\n",
       "      <td>426000000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>100</th>\n",
       "      <td>Braga</td>\n",
       "      <td>FC Porto</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>A</td>\n",
       "      <td>1503861300</td>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>...</td>\n",
       "      <td>75</td>\n",
       "      <td>74</td>\n",
       "      <td>79</td>\n",
       "      <td>78</td>\n",
       "      <td>79</td>\n",
       "      <td>79</td>\n",
       "      <td>1562.021240</td>\n",
       "      <td>1745.115723</td>\n",
       "      <td>5.740000e+07</td>\n",
       "      <td>177650000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1000</th>\n",
       "      <td>Red Star FC 93</td>\n",
       "      <td>Strasbourg</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>D</td>\n",
       "      <td>1487358000</td>\n",
       "      <td>16</td>\n",
       "      <td>6</td>\n",
       "      <td>26</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>66</td>\n",
       "      <td>67</td>\n",
       "      <td>66</td>\n",
       "      <td>68</td>\n",
       "      <td>67</td>\n",
       "      <td>66</td>\n",
       "      <td>1407.163086</td>\n",
       "      <td>1422.037231</td>\n",
       "      <td>1.210000e+07</td>\n",
       "      <td>9850000.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 30 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "           home_name   away_name  home_score  away_score final_result  \\\n",
       "0           FC Porto    Nacional           3           1            H   \n",
       "1           Tenerife       Elche           2           1            H   \n",
       "10          Man City     Everton           3           1            H   \n",
       "100            Braga    FC Porto           0           1            A   \n",
       "1000  Red Star FC 93  Strasbourg           1           1            D   \n",
       "\n",
       "            time  home_pos  away_pos  round  home_last5all_home  \\\n",
       "0     1546895700         1        11     16                   9   \n",
       "1     1546632000        18        15     20                   2   \n",
       "10    1544877000         2         7     17                  16   \n",
       "100   1503861300         6         3      4                   3   \n",
       "1000  1487358000        16         6     26                   0   \n",
       "\n",
       "           ...        fifa_home_mid  fifa_home_def  fifa_away_ova  \\\n",
       "0          ...                   81             80             70   \n",
       "1          ...                   71             69             67   \n",
       "10         ...                   88             83             79   \n",
       "100        ...                   75             74             79   \n",
       "1000       ...                   66             67             66   \n",
       "\n",
       "      fifa_away_att  fifa_away_mid  fifa_away_def  elo_home_score  \\\n",
       "0                70             70             69     1813.906738   \n",
       "1                69             67             67     1513.179932   \n",
       "10               79             80             79     1986.672974   \n",
       "100              78             79             79     1562.021240   \n",
       "1000             68             67             66     1407.163086   \n",
       "\n",
       "      elo_away_score  tfm_value_home  tfm_value_away  \n",
       "0        1392.986084    2.779800e+08       9450000.0  \n",
       "1        1475.512939    1.090000e+07       9600000.0  \n",
       "10       1699.472778    1.090000e+09     426000000.0  \n",
       "100      1745.115723    5.740000e+07     177650000.0  \n",
       "1000     1422.037231    1.210000e+07       9850000.0  \n",
       "\n",
       "[5 rows x 30 columns]"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 5163 entries, 0 to 999\n",
      "Data columns (total 30 columns):\n",
      "home_name                  5163 non-null object\n",
      "away_name                  5163 non-null object\n",
      "home_score                 5163 non-null int64\n",
      "away_score                 5163 non-null int64\n",
      "final_result               5163 non-null object\n",
      "time                       5163 non-null int64\n",
      "home_pos                   5163 non-null int64\n",
      "away_pos                   5163 non-null int64\n",
      "round                      5163 non-null int64\n",
      "home_last5all_home         5163 non-null int64\n",
      "home_last5all_home_win     5163 non-null int64\n",
      "home_last5all_home_draw    5163 non-null int64\n",
      "home_last5all_home_lose    5163 non-null int64\n",
      "away_last5all_away         5163 non-null int64\n",
      "away_last5all_away_win     5163 non-null int64\n",
      "away_last5all_away_draw    5163 non-null int64\n",
      "away_last5all_away_lose    5163 non-null int64\n",
      "last5all_home_away_dif     5163 non-null int64\n",
      "fifa_home_ova              5163 non-null int64\n",
      "fifa_home_att              5163 non-null int64\n",
      "fifa_home_mid              5163 non-null int64\n",
      "fifa_home_def              5163 non-null int64\n",
      "fifa_away_ova              5163 non-null int64\n",
      "fifa_away_att              5163 non-null int64\n",
      "fifa_away_mid              5163 non-null int64\n",
      "fifa_away_def              5163 non-null int64\n",
      "elo_home_score             5163 non-null float64\n",
      "elo_away_score             5163 non-null float64\n",
      "tfm_value_home             5163 non-null float64\n",
      "tfm_value_away             5163 non-null float64\n",
      "dtypes: float64(4), int64(23), object(3)\n",
      "memory usage: 1.2+ MB\n"
     ]
    }
   ],
   "source": [
    "df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tfm_value_away             0\n",
       "tfm_value_home             0\n",
       "away_name                  0\n",
       "home_score                 0\n",
       "away_score                 0\n",
       "final_result               0\n",
       "time                       0\n",
       "home_pos                   0\n",
       "away_pos                   0\n",
       "round                      0\n",
       "home_last5all_home         0\n",
       "home_last5all_home_win     0\n",
       "home_last5all_home_draw    0\n",
       "home_last5all_home_lose    0\n",
       "away_last5all_away         0\n",
       "away_last5all_away_win     0\n",
       "away_last5all_away_draw    0\n",
       "away_last5all_away_lose    0\n",
       "last5all_home_away_dif     0\n",
       "fifa_home_ova              0\n",
       "fifa_home_att              0\n",
       "fifa_home_mid              0\n",
       "fifa_home_def              0\n",
       "fifa_away_ova              0\n",
       "fifa_away_att              0\n",
       "fifa_away_mid              0\n",
       "fifa_away_def              0\n",
       "elo_home_score             0\n",
       "elo_away_score             0\n",
       "home_name                  0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Verificando se existe algum dado ausente\n",
    "df.isnull().sum().sort_values(ascending=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0xc78914b630>"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.heatmap(df.isna(), yticklabels=False, cbar=False, cmap='Paired')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "H    2306\n",
       "A    1500\n",
       "D    1357\n",
       "Name: final_result, dtype: int64"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Verificando quantos exemplos há no dataset da classe alvo\n",
    "df['final_result'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<seaborn.axisgrid.FacetGrid at 0xc78912b080>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.catplot('final_result',data=df, kind='count')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1472, 30)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ds1 = df[(df['tfm_value_home'] > df['tfm_value_away'])]\n",
    "ds1 = ds1[ds1['final_result']==\"H\"]\n",
    "ds1.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1027, 30)"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ds2 = df[(df['tfm_value_home'] < df['tfm_value_away'])]\n",
    "ds2 = ds2[ds2['final_result']==\"A\"]\n",
    "ds2.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(470, 30)"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ds3 = df[(df['tfm_value_home'] > df['tfm_value_away'])]\n",
    "ds3 = ds3[ds3['final_result']==\"A\"]\n",
    "ds3.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(831, 30)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ds4 = df[(df['tfm_value_home'] < df['tfm_value_away'])]\n",
    "ds4 = ds4[ds4['final_result']==\"H\"]\n",
    "ds4.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0xc78911ceb8>"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "conj = {'valor_mercado': ['Maior valor de mercado', 'Menor valor de mercado'], 'quantidade': [ds1.shape[0] + ds2.shape[0], ds3.shape[0] + ds4.shape[0]]}\n",
    "dfH1 = pd.DataFrame(conj, columns=['valor_mercado', 'quantidade'])\n",
    "sns.barplot(x='valor_mercado', y='quantidade', data=dfH1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h2>Criação do Modelo</h2>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Primeira Versão de Features<br/>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "X = df.drop(['home_name', 'away_name', 'home_score', 'away_score', 'time', 'final_result'], inplace=False, axis=1)\n",
    "y = df['final_result']\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=123)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h5>Gradient Boosting</h5>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "pipe = Pipeline(steps= [(\"standardscaler\", StandardScaler()),\n",
    "                        (\"gradientboostingclassifier\", GradientBoostingClassifier())\n",
    "                       ])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "param_grid_GB = {\n",
    "    \"gradientboostingclassifier__learning_rate\": [0.0001, 0.1, 0.2],\n",
    "    \"gradientboostingclassifier__max_depth\":[3, 5],\n",
    "    \"gradientboostingclassifier__max_features\":[\"log2\",\"sqrt\"],\n",
    "    \"gradientboostingclassifier__n_estimators\":[10, 100, 300]\n",
    "    }"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "grid = GridSearchCV(pipe, param_grid=param_grid_GB, scoring=\"accuracy\", iid=False, refit=True, cv=5, return_train_score=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GridSearchCV(cv=5, error_score='raise-deprecating',\n",
       "       estimator=Pipeline(memory=None,\n",
       "     steps=[('standardscaler', StandardScaler(copy=True, with_mean=True, with_std=True)), ('gradientboostingclassifier', GradientBoostingClassifier(criterion='friedman_mse', init=None,\n",
       "              learning_rate=0.1, loss='deviance', max_depth=3,\n",
       "              max_features=None, max_leaf_nodes=None,\n",
       "   ...    subsample=1.0, tol=0.0001, validation_fraction=0.1,\n",
       "              verbose=0, warm_start=False))]),\n",
       "       fit_params=None, iid=False, n_jobs=None,\n",
       "       param_grid={'gradientboostingclassifier__learning_rate': [0.0001, 0.1, 0.2], 'gradientboostingclassifier__max_depth': [3, 5], 'gradientboostingclassifier__max_features': ['log2', 'sqrt'], 'gradientboostingclassifier__n_estimators': [10, 100, 300]},\n",
       "       pre_dispatch='2*n_jobs', refit=True, return_train_score=False,\n",
       "       scoring='accuracy', verbose=0)"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "grid.fit(X,y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Pipeline(memory=None,\n",
      "     steps=[('standardscaler', StandardScaler(copy=True, with_mean=True, with_std=True)), ('gradientboostingclassifier', GradientBoostingClassifier(criterion='friedman_mse', init=None,\n",
      "              learning_rate=0.1, loss='deviance', max_depth=5,\n",
      "              max_features='sqrt', max_leaf_nodes=None,\n",
      " ...    subsample=1.0, tol=0.0001, validation_fraction=0.1,\n",
      "              verbose=0, warm_start=False))])\n",
      "{'gradientboostingclassifier__learning_rate': 0.1, 'gradientboostingclassifier__max_depth': 5, 'gradientboostingclassifier__max_features': 'sqrt', 'gradientboostingclassifier__n_estimators': 10}\n",
      "0.5097796075293725\n"
     ]
    }
   ],
   "source": [
    "print(grid.best_estimator_)\n",
    "print(grid.best_params_)\n",
    "print(grid.best_score_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h5>Random Forest</h5>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "pipeRF = Pipeline(steps= [(\"standardscaler\", StandardScaler()),\n",
    "                        (\"randomforestclassifier\", RandomForestClassifier())\n",
    "                       ])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "param_grid_RF = { \n",
    "    'randomforestclassifier__n_estimators': [10, 100, 500],\n",
    "    'randomforestclassifier__max_features': ['sqrt', 'log2'],\n",
    "    'randomforestclassifier__max_depth' : [4, 8],\n",
    "    'randomforestclassifier__min_samples_split': [2, 8],\n",
    "    'randomforestclassifier__min_samples_leaf': [1, 5]\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "gridRF = GridSearchCV(pipeRF, param_grid=param_grid_RF, scoring='accuracy', iid=False, refit=True, cv=5, return_train_score=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GridSearchCV(cv=5, error_score='raise-deprecating',\n",
       "       estimator=Pipeline(memory=None,\n",
       "     steps=[('standardscaler', StandardScaler(copy=True, with_mean=True, with_std=True)), ('randomforestclassifier', RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',\n",
       "            max_depth=None, max_features='auto', max_leaf_nodes=None,\n",
       "            min_impurity_decrease=0.0, min...obs=None,\n",
       "            oob_score=False, random_state=None, verbose=0,\n",
       "            warm_start=False))]),\n",
       "       fit_params=None, iid=False, n_jobs=None,\n",
       "       param_grid={'randomforestclassifier__n_estimators': [10, 100, 500], 'randomforestclassifier__max_features': ['sqrt', 'log2'], 'randomforestclassifier__max_depth': [4, 8], 'randomforestclassifier__min_samples_split': [2, 8], 'randomforestclassifier__min_samples_leaf': [1, 5]},\n",
       "       pre_dispatch='2*n_jobs', refit=True, return_train_score=False,\n",
       "       scoring='accuracy', verbose=0)"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gridRF.fit(X,y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(make_scorer(accuracy_score), Pipeline(memory=None,\n",
       "      steps=[('standardscaler', StandardScaler(copy=True, with_mean=True, with_std=True)), ('randomforestclassifier', RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',\n",
       "             max_depth=8, max_features='log2', max_leaf_nodes=None,\n",
       "             min_impurity_decrease=0.0, min_im...obs=None,\n",
       "             oob_score=False, random_state=None, verbose=0,\n",
       "             warm_start=False))]), {'randomforestclassifier__max_depth': 8,\n",
       "  'randomforestclassifier__max_features': 'log2',\n",
       "  'randomforestclassifier__min_samples_leaf': 5,\n",
       "  'randomforestclassifier__min_samples_split': 2,\n",
       "  'randomforestclassifier__n_estimators': 500}, 0.5136536997218141)"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gridRF.scorer_, gridRF.best_estimator_, gridRF.best_params_, gridRF.best_score_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "standard_scaler = StandardScaler()\n",
    "X_train_std = standard_scaler.fit_transform(X_train)\n",
    "X_test_std = standard_scaler.transform(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',\n",
       "            max_depth=None, max_features='auto', max_leaf_nodes=None,\n",
       "            min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "            min_samples_leaf=1, min_samples_split=2,\n",
       "            min_weight_fraction_leaf=0.0, n_estimators=500, n_jobs=None,\n",
       "            oob_score=False, random_state=None, verbose=0,\n",
       "            warm_start=False)"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rf = RandomForestClassifier(n_estimators=500)\n",
    "rf.fit(X_train_std, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "predRF = rf.predict(X_test_std)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[166  47 251]\n",
      " [ 95  43 279]\n",
      " [ 84  53 531]]\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           A       0.48      0.36      0.41       464\n",
      "           D       0.30      0.10      0.15       417\n",
      "           H       0.50      0.79      0.61       668\n",
      "\n",
      "   micro avg       0.48      0.48      0.48      1549\n",
      "   macro avg       0.43      0.42      0.39      1549\n",
      "weighted avg       0.44      0.48      0.43      1549\n",
      "\n",
      "0.4777275661717237\n"
     ]
    }
   ],
   "source": [
    "print(confusion_matrix(y_test, predRF))\n",
    "print(classification_report(y_test, predRF))\n",
    "print(accuracy_score(y_test, predRF))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Verifica a importância de cada variável no modelo\n",
    "fName = pd.DataFrame(X.columns, columns=['Feature'])\n",
    "fImp = pd.DataFrame(rf.feature_importances_, columns=['Importance'])\n",
    "dfFI = pd.concat([fName, fImp], axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Feature</th>\n",
       "      <th>Importance</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>elo_away_score</td>\n",
       "      <td>0.074450</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>elo_home_score</td>\n",
       "      <td>0.071911</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>tfm_value_away</td>\n",
       "      <td>0.065403</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>tfm_value_home</td>\n",
       "      <td>0.065318</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>round</td>\n",
       "      <td>0.056260</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>home_pos</td>\n",
       "      <td>0.049481</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>away_pos</td>\n",
       "      <td>0.048713</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>last5all_home_away_dif</td>\n",
       "      <td>0.047333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>away_last5all_away</td>\n",
       "      <td>0.043882</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>home_last5all_home</td>\n",
       "      <td>0.043132</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>fifa_home_att</td>\n",
       "      <td>0.041260</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>fifa_away_att</td>\n",
       "      <td>0.040300</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>fifa_home_mid</td>\n",
       "      <td>0.039271</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>fifa_away_mid</td>\n",
       "      <td>0.038987</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>fifa_away_def</td>\n",
       "      <td>0.037927</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>fifa_home_def</td>\n",
       "      <td>0.036825</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>fifa_away_ova</td>\n",
       "      <td>0.034238</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>fifa_home_ova</td>\n",
       "      <td>0.032972</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>home_last5all_home_win</td>\n",
       "      <td>0.023221</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>away_last5all_away_lose</td>\n",
       "      <td>0.022645</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>away_last5all_away_draw</td>\n",
       "      <td>0.022451</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>home_last5all_home_draw</td>\n",
       "      <td>0.022297</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>away_last5all_away_win</td>\n",
       "      <td>0.021126</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>home_last5all_home_lose</td>\n",
       "      <td>0.020597</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                    Feature  Importance\n",
       "21           elo_away_score    0.074450\n",
       "20           elo_home_score    0.071911\n",
       "23           tfm_value_away    0.065403\n",
       "22           tfm_value_home    0.065318\n",
       "2                     round    0.056260\n",
       "0                  home_pos    0.049481\n",
       "1                  away_pos    0.048713\n",
       "11   last5all_home_away_dif    0.047333\n",
       "7        away_last5all_away    0.043882\n",
       "3        home_last5all_home    0.043132\n",
       "13            fifa_home_att    0.041260\n",
       "17            fifa_away_att    0.040300\n",
       "14            fifa_home_mid    0.039271\n",
       "18            fifa_away_mid    0.038987\n",
       "19            fifa_away_def    0.037927\n",
       "15            fifa_home_def    0.036825\n",
       "16            fifa_away_ova    0.034238\n",
       "12            fifa_home_ova    0.032972\n",
       "4    home_last5all_home_win    0.023221\n",
       "10  away_last5all_away_lose    0.022645\n",
       "9   away_last5all_away_draw    0.022451\n",
       "5   home_last5all_home_draw    0.022297\n",
       "8    away_last5all_away_win    0.021126\n",
       "6   home_last5all_home_lose    0.020597"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dfFI.sort_values('Importance', ascending=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "dfIF = dfFI.sort_values('Importance', ascending=False)\n",
    "\n",
    "fIF = dfIF['Feature']\n",
    "iIF = dfIF['Importance']\n",
    "\n",
    "plt.title('Importância dos Atributos')\n",
    "plt.barh(range(len(fIF)), iIF, color='royalblue', align='center')\n",
    "plt.yticks(range(len(fIF)), fIF)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "features = list(dfFI[dfFI['Importance'] < 0.045]['Feature'])\n",
    "features.append('home_name')\n",
    "features.append('away_name')\n",
    "features.append('home_score')\n",
    "features.append('away_score')\n",
    "features.append('time')\n",
    "features.append('final_result')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h6>Pré-processamento</h6>\n",
    "Redução de dimensionalidade, com base nas features mais importantes, escolhido pelo Random Forest<br/>\n",
    "Criação de novas dimensões para verificar se há uma melhora no aprendizado"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "X = df.drop(features, inplace=False, axis=1)\n",
    "X['score_elo_home_away_dif'] = X['elo_home_score'] - X['elo_away_score']\n",
    "X['tfm_value_home_away_dif'] = X['tfm_value_home'] - X['tfm_value_away']\n",
    "y = df['final_result']\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['home_pos', 'away_pos', 'round', 'last5all_home_away_dif',\n",
       "       'elo_home_score', 'elo_away_score', 'tfm_value_home', 'tfm_value_away',\n",
       "       'score_elo_home_away_dif', 'tfm_value_home_away_dif'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GridSearchCV(cv=5, error_score='raise-deprecating',\n",
       "       estimator=Pipeline(memory=None,\n",
       "     steps=[('standardscaler', StandardScaler(copy=True, with_mean=True, with_std=True)), ('gradientboostingclassifier', GradientBoostingClassifier(criterion='friedman_mse', init=None,\n",
       "              learning_rate=0.1, loss='deviance', max_depth=3,\n",
       "              max_features=None, max_leaf_nodes=None,\n",
       "   ...    subsample=1.0, tol=0.0001, validation_fraction=0.1,\n",
       "              verbose=0, warm_start=False))]),\n",
       "       fit_params=None, iid=False, n_jobs=None,\n",
       "       param_grid={'gradientboostingclassifier__learning_rate': [0.0001, 0.1, 0.2], 'gradientboostingclassifier__max_depth': [3, 5], 'gradientboostingclassifier__max_features': ['log2', 'sqrt'], 'gradientboostingclassifier__n_estimators': [10, 100, 300]},\n",
       "       pre_dispatch='2*n_jobs', refit=True, return_train_score=False,\n",
       "       scoring='accuracy', verbose=0)"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "grid.fit(X,y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Pipeline(memory=None,\n",
      "     steps=[('standardscaler', StandardScaler(copy=True, with_mean=True, with_std=True)), ('gradientboostingclassifier', GradientBoostingClassifier(criterion='friedman_mse', init=None,\n",
      "              learning_rate=0.2, loss='deviance', max_depth=3,\n",
      "              max_features='log2', max_leaf_nodes=None,\n",
      " ...    subsample=1.0, tol=0.0001, validation_fraction=0.1,\n",
      "              verbose=0, warm_start=False))])\n",
      "{'gradientboostingclassifier__learning_rate': 0.2, 'gradientboostingclassifier__max_depth': 3, 'gradientboostingclassifier__max_features': 'log2', 'gradientboostingclassifier__n_estimators': 10}\n",
      "0.5159781561984855\n"
     ]
    }
   ],
   "source": [
    "print(grid.best_estimator_)\n",
    "print(grid.best_params_)\n",
    "print(grid.best_score_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h6>Gravar o modelo criado em disco</h6>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Salva o modelo em disco\n",
    "filename = 'modeloGB.sav' \n",
    "dump(grid, open(filename, 'wb'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h6>SVM</h6>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "svc = SVC(C=5.0, kernel='rbf', degree=5, probability=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "SVC(C=5.0, cache_size=200, class_weight=None, coef0=0.0,\n",
       "  decision_function_shape='ovr', degree=5, gamma='auto_deprecated',\n",
       "  kernel='rbf', max_iter=-1, probability=True, random_state=None,\n",
       "  shrinking=True, tol=0.001, verbose=False)"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "svc.fit(X_train_std, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred = svc.predict_proba(X_test_std)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.29881325, 0.25529595, 0.4458908 ],\n",
       "       [0.29330926, 0.26481083, 0.4418799 ],\n",
       "       [0.29584689, 0.25538871, 0.4487644 ],\n",
       "       ...,\n",
       "       [0.28403515, 0.26098479, 0.45498006],\n",
       "       [0.28884843, 0.26740694, 0.44374463],\n",
       "       [0.2967101 , 0.26045221, 0.44283769]])"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pred[:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['A', 'D', 'H'], dtype=object)"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "svc.classes_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h2>Conclusão</h2>\n",
    "* O modelo não conseguiu generalizar os dados (Overfitting)\n",
    "* Não consegue encontrar um padrão nos exemplos de empate \n",
    "* Possíveis soluções:\n",
    "    + Estudar mais o cenário e regra de negócio para criar atributos relevantes\n",
    "    + Capturar mais exemplos de partidas de futebol (principalmente os empates)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}