Notebook/Final_model_VKaggle.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "5a660d81",
   "metadata": {
    "papermill": {
     "duration": 0.023016,
     "end_time": "2021-10-17T06:35:29.169259",
     "exception": false,
     "start_time": "2021-10-17T06:35:29.146243",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Olá!\n",
    "\n",
    "Este é o meu notebook e é fruto do trabalho conjunto dos cientistas de dados Marcos Mansur e Thiago Ouverney para a competição VLabs Data Challenge, onde deseja-se prever o LTV dos próximos 90 dias aberto por cliente a partir de 14 meses de dados de compras em diferentes canais. \n",
    "\n",
    "Esse notebook retrata o modelo com melhor score que fizemos."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "237eb467",
   "metadata": {
    "papermill": {
     "duration": 0.020732,
     "end_time": "2021-10-17T06:35:29.213081",
     "exception": false,
     "start_time": "2021-10-17T06:35:29.192349",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Carregar Libs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "f19a66bd",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-10-17T06:35:29.263601Z",
     "iopub.status.busy": "2021-10-17T06:35:29.262939Z",
     "iopub.status.idle": "2021-10-17T06:35:30.327295Z",
     "shell.execute_reply": "2021-10-17T06:35:30.326076Z",
     "shell.execute_reply.started": "2021-10-17T05:40:19.488213Z"
    },
    "papermill": {
     "duration": 1.093408,
     "end_time": "2021-10-17T06:35:30.327476",
     "exception": false,
     "start_time": "2021-10-17T06:35:29.234068",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.model_selection import cross_val_score, train_test_split, learning_curve, GridSearchCV\n",
    "from sklearn.metrics import mean_squared_error\n",
    "from sklearn.linear_model import LinearRegression, ElasticNet\n",
    "from sklearn.base import BaseEstimator, TransformerMixin\n",
    "from sklearn.preprocessing import FunctionTransformer\n",
    "from sklearn.pipeline import make_pipeline\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "589b53de",
   "metadata": {
    "papermill": {
     "duration": 0.020526,
     "end_time": "2021-10-17T06:35:30.368660",
     "exception": false,
     "start_time": "2021-10-17T06:35:30.348134",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Carregar dados"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "b0b148e6",
   "metadata": {
    "_cell_guid": "b1076dfc-b9ad-4769-8c92-a6c4dae69d19",
    "_uuid": "8f2839f25d086af736a60e9eeb907d3b93b6e0e5",
    "execution": {
     "iopub.execute_input": "2021-10-17T06:35:30.414791Z",
     "iopub.status.busy": "2021-10-17T06:35:30.414019Z",
     "iopub.status.idle": "2021-10-17T06:35:30.420855Z",
     "shell.execute_reply": "2021-10-17T06:35:30.421484Z",
     "shell.execute_reply.started": "2021-10-17T04:34:13.083098Z"
    },
    "papermill": {
     "duration": 0.032652,
     "end_time": "2021-10-17T06:35:30.421655",
     "exception": false,
     "start_time": "2021-10-17T06:35:30.389003",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/kaggle/input/VLabs-DC/sales_20_21_train.csv\n",
      "/kaggle/input/VLabs-DC/sample_submission.csv\n"
     ]
    }
   ],
   "source": [
    "# This Python 3 environment comes with many helpful analytics libraries installed\n",
    "# It is defined by the kaggle/python Docker image: https://github.com/kaggle/docker-python\n",
    "\n",
    "import os\n",
    "for dirname, _, filenames in os.walk('/kaggle/input'):\n",
    "    for filename in filenames:\n",
    "        print(os.path.join(dirname, filename))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "7166057c",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-10-17T06:35:30.467289Z",
     "iopub.status.busy": "2021-10-17T06:35:30.466324Z",
     "iopub.status.idle": "2021-10-17T06:35:32.382031Z",
     "shell.execute_reply": "2021-10-17T06:35:32.381515Z",
     "shell.execute_reply.started": "2021-10-17T04:36:28.921212Z"
    },
    "papermill": {
     "duration": 1.939581,
     "end_time": "2021-10-17T06:35:32.382205",
     "exception": false,
     "start_time": "2021-10-17T06:35:30.442624",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "df = pd.read_csv('/kaggle/input/VLabs-DC/sales_20_21_train.csv', header=0, parse_dates=['DT_VENDA'])\n",
    "# df = pd.read_csv('https://raw.githubusercontent.com/marcos-mansur/vlabs-challenge/main/Data/sales_20_21_train.csv', header=0, parse_dates=['DT_VENDA'])\n",
    "df_sub = pd.read_csv('/kaggle/input/VLabs-DC/sample_submission.csv',header=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "febe3bb3",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-10-17T06:35:32.431449Z",
     "iopub.status.busy": "2021-10-17T06:35:32.430774Z",
     "iopub.status.idle": "2021-10-17T06:35:32.443825Z",
     "shell.execute_reply": "2021-10-17T06:35:32.444313Z",
     "shell.execute_reply.started": "2021-10-17T06:08:27.953885Z"
    },
    "papermill": {
     "duration": 0.041557,
     "end_time": "2021-10-17T06:35:32.444488",
     "exception": false,
     "start_time": "2021-10-17T06:35:32.402931",
     "status": "completed"
    },
    "tags": []
   },
   "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_VENDA</th>\n",
       "      <th>DT_VENDA</th>\n",
       "      <th>LOJA</th>\n",
       "      <th>QTD_SKU</th>\n",
       "      <th>VALOR</th>\n",
       "      <th>ID_CLIENTE</th>\n",
       "      <th>CANAL</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1176550</td>\n",
       "      <td>2020-01-02</td>\n",
       "      <td>37</td>\n",
       "      <td>1</td>\n",
       "      <td>9.50</td>\n",
       "      <td>337763</td>\n",
       "      <td>FIS</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>11683461</td>\n",
       "      <td>2020-01-02</td>\n",
       "      <td>15</td>\n",
       "      <td>1</td>\n",
       "      <td>6.64</td>\n",
       "      <td>195687</td>\n",
       "      <td>FIS</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3326484</td>\n",
       "      <td>2020-01-02</td>\n",
       "      <td>29</td>\n",
       "      <td>5</td>\n",
       "      <td>446.87</td>\n",
       "      <td>56059</td>\n",
       "      <td>FIS</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>7317475</td>\n",
       "      <td>2020-01-02</td>\n",
       "      <td>18</td>\n",
       "      <td>1</td>\n",
       "      <td>6.13</td>\n",
       "      <td>297533</td>\n",
       "      <td>FIS</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4366938</td>\n",
       "      <td>2020-01-02</td>\n",
       "      <td>23</td>\n",
       "      <td>1</td>\n",
       "      <td>13.17</td>\n",
       "      <td>181284</td>\n",
       "      <td>FIS</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   ID_VENDA   DT_VENDA  LOJA  QTD_SKU   VALOR  ID_CLIENTE CANAL\n",
       "0   1176550 2020-01-02    37        1    9.50      337763   FIS\n",
       "1  11683461 2020-01-02    15        1    6.64      195687   FIS\n",
       "2   3326484 2020-01-02    29        5  446.87       56059   FIS\n",
       "3   7317475 2020-01-02    18        1    6.13      297533   FIS\n",
       "4   4366938 2020-01-02    23        1   13.17      181284   FIS"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "47f874fa",
   "metadata": {
    "papermill": {
     "duration": 0.020857,
     "end_time": "2021-10-17T06:35:32.486721",
     "exception": false,
     "start_time": "2021-10-17T06:35:32.465864",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Funções"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "88bea879",
   "metadata": {
    "papermill": {
     "duration": 0.020903,
     "end_time": "2021-10-17T06:35:32.531576",
     "exception": false,
     "start_time": "2021-10-17T06:35:32.510673",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Plot curvas de aprendizado"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "e18e2df2",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-10-17T06:35:32.576569Z",
     "iopub.status.busy": "2021-10-17T06:35:32.575879Z",
     "iopub.status.idle": "2021-10-17T06:35:32.585559Z",
     "shell.execute_reply": "2021-10-17T06:35:32.586075Z",
     "shell.execute_reply.started": "2021-10-17T05:40:55.852792Z"
    },
    "papermill": {
     "duration": 0.033832,
     "end_time": "2021-10-17T06:35:32.586271",
     "exception": false,
     "start_time": "2021-10-17T06:35:32.552439",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "def plot_learning_curve(estimator, title, X, y, axes=None, ylim=None, cv=None,\n",
    "                        n_jobs=None, train_sizes=np.linspace(.1, 1.0, 5)):\n",
    "    \"\"\"\n",
    "    Generate plot: learning curve\n",
    "\n",
    "    \"\"\"\n",
    "    if axes is None:\n",
    "        _, axes = plt.subplots(1, figsize=(20, 10))\n",
    "\n",
    "    axes.set_title(title)\n",
    "    if ylim is not None:\n",
    "        axes.set_ylim(*ylim)\n",
    "    axes.set_xlabel(\"Training examples\")\n",
    "    axes.set_ylabel(\"Score\")\n",
    "\n",
    "    train_sizes, train_scores, test_scores, fit_times, _ = \\\n",
    "        learning_curve(estimator, X, y, cv=cv, n_jobs=n_jobs,\n",
    "                       train_sizes=train_sizes, scoring='neg_root_mean_squared_error',\n",
    "                       return_times=True)\n",
    "    train_scores_mean = np.mean(train_scores, axis=1)\n",
    "    train_scores_std = np.std(train_scores, axis=1)\n",
    "    test_scores_mean = np.mean(test_scores, axis=1)\n",
    "    test_scores_std = np.std(test_scores, axis=1)\n",
    "    fit_times_mean = np.mean(fit_times, axis=1)\n",
    "    fit_times_std = np.std(fit_times, axis=1)\n",
    "\n",
    "    # Plot learning curve\n",
    "    axes.grid()\n",
    "    # plot intervalos de spread \n",
    "    axes.fill_between(train_sizes, train_scores_mean - train_scores_std,\n",
    "                         train_scores_mean + train_scores_std, alpha=0.1,\n",
    "                         color=\"r\")\n",
    "    axes.fill_between(train_sizes, test_scores_mean - test_scores_std,\n",
    "                         test_scores_mean + test_scores_std, alpha=0.1,\n",
    "                         color=\"g\")\n",
    "    # curvas\n",
    "    axes.plot(train_sizes, train_scores_mean, 'o-', color=\"r\",\n",
    "                 label=\"Training score\")\n",
    "    axes.plot(train_sizes, test_scores_mean, 'o-', color=\"g\",\n",
    "                 label=\"Cross-validation score\")\n",
    "    axes.legend(loc=\"best\")\n",
    "    return plt"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "577c71b7",
   "metadata": {
    "papermill": {
     "duration": 0.020607,
     "end_time": "2021-10-17T06:35:32.628013",
     "exception": false,
     "start_time": "2021-10-17T06:35:32.607406",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Pré-processamento"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "5c4b01ad",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-10-17T06:35:32.672741Z",
     "iopub.status.busy": "2021-10-17T06:35:32.672053Z",
     "iopub.status.idle": "2021-10-17T06:35:32.686793Z",
     "shell.execute_reply": "2021-10-17T06:35:32.687344Z",
     "shell.execute_reply.started": "2021-10-17T05:46:30.929113Z"
    },
    "papermill": {
     "duration": 0.038715,
     "end_time": "2021-10-17T06:35:32.687520",
     "exception": false,
     "start_time": "2021-10-17T06:35:32.648805",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "def split_per_dates(df,train_start,target_start, sub_start):\n",
    "    \" divide os dados de treino, validação e submissão por data\"\n",
    "    # target do treinamento\n",
    "    df_target = df[df.DT_VENDA > target_start]\n",
    "    y_train = df_target.groupby('ID_CLIENTE')['VALOR'].sum()\n",
    "    # dados de treinamento\n",
    "    df_train = df[(df['DT_VENDA']>train_start) & (df['DT_VENDA']<target_start)].copy()\n",
    "    # dados para previsão\n",
    "    df_test = df[df['DT_VENDA']>sub_start]\n",
    "    return df_train, y_train, df_test\n",
    "\n",
    "def treated_data(df_pp):\n",
    "    \" transforma o dataset aberto por compras ordenado no tempo \"\n",
    "    \" em um dataset por cliente e cria features úteis\"\n",
    "    \n",
    "    # intervalos de tempo\n",
    "    delta90 = pd.to_timedelta(90,unit='d')\n",
    "    delta120 = pd.to_timedelta(120,unit='d')\n",
    "    delta150 = pd.to_timedelta(150,unit='d')\n",
    "    delta180 = pd.to_timedelta(180,unit='d')\n",
    "    delta360 = pd.to_timedelta(360,unit='d')\n",
    "    delta270 = pd.to_timedelta(270,unit='d')\n",
    "\n",
    "    DICT_AUX_DATAS = {'90':delta90,\n",
    "                      '120':delta120,\n",
    "                      '150':delta150,\n",
    "                      '180':delta180}\n",
    "\n",
    "    max_date = df_pp['DT_VENDA'].max()\n",
    "    \n",
    "    #agrupar por cliente somando o valor\n",
    "    cust_revenue = df_pp.groupby(['ID_CLIENTE'])['VALOR'].sum().copy()\n",
    "    # transformar em df\n",
    "    cust_rev = pd.DataFrame(cust_revenue)\n",
    "    # feature com número de compras feitas no período de teste\n",
    "    cust_rev['Frequency'] = df_pp.groupby(['ID_CLIENTE'])['ID_VENDA'].count()\n",
    "\n",
    "\n",
    "    # media entre os valores das compras de cada cliente\n",
    "    cust_rev['valor medio'] = df_pp.groupby('ID_CLIENTE')['VALOR'].mean()\n",
    "    # mediana\n",
    "    cust_rev['valor mediana'] = df_pp.groupby('ID_CLIENTE')['VALOR'].median()\n",
    "    # desvio padrao\n",
    "    cust_rev['valor desvio'] = df_pp.groupby('ID_CLIENTE')['VALOR'].std()\n",
    "    \n",
    "    \n",
    "    # data da ultima compra\n",
    "    cust_rev['Recency'] = max_date - df_pp.groupby(['ID_CLIENTE'])['DT_VENDA'].max()\n",
    "    cust_rev['Recency'] = cust_rev['Recency'].apply(lambda x: x.days)\n",
    "    \n",
    "    # soma e média do VALOR de diferentes períodos de cada cliente    \n",
    "    for item, valor in DICT_AUX_DATAS.items():\n",
    "        cust_rev[f'valor_{item}m'] = df_pp[\n",
    "              df_pp['DT_VENDA']>(df_pp['DT_VENDA'].max() - valor)].groupby(['ID_CLIENTE'])['VALOR'].sum()\n",
    "        cust_rev[f'valormedio_{item}m'] = df_pp[\n",
    "              df_pp['DT_VENDA']>(df_pp['DT_VENDA'].max() - valor)].groupby(['ID_CLIENTE'])['VALOR'].mean()\n",
    "\n",
    "    # soma do valor de 3 meses há 1 ano atrás \n",
    "    ref = df_pp['DT_VENDA'].max()\n",
    "    cust_rev[f'valorsum_1ano_atras_m'] = df_pp[\n",
    "            (df_pp['DT_VENDA'] < (ref - delta270) ) &\n",
    "           (df_pp['DT_VENDA'] > (ref - delta360) )\n",
    "            ].groupby(['ID_CLIENTE'])['VALOR'].sum()\n",
    "            \n",
    "    return cust_rev.fillna(0)\n",
    "\n",
    "def transform_df(df):\n",
    "    df = df.copy()\n",
    "    \n",
    "    # divide os dados de treino e previsão pela data\n",
    "    df_train, target_train, df_test = split_per_dates(df,train_start,target_start,sub_start)\n",
    "    # modifica o df e cria features nos dados de treino\n",
    "    df_train1 = treated_data(df_train)\n",
    "    # para alinhar o index do target do treino com os dados de treino\n",
    "    df_train2 = df_train1.join(other=target_train, on='ID_CLIENTE', lsuffix='_sum', rsuffix='_TARGET')\n",
    "    x_train = df_train2.drop('VALOR_TARGET',axis=1)\n",
    "    y_train = df_train2.VALOR_TARGET.fillna(0)\n",
    "    \n",
    "    # modifica o df e cria features nos dados de test\n",
    "    x_test = treated_data(df_test)\n",
    "    return x_train, y_train, x_test"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1c257187",
   "metadata": {
    "papermill": {
     "duration": 0.020737,
     "end_time": "2021-10-17T06:35:32.729330",
     "exception": false,
     "start_time": "2021-10-17T06:35:32.708593",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Estratégia\n",
    "\n",
    "Primeiro, vamos dividir os dados por data em x de treino (período dos dados para treinar o modelo), y de treino (o período do target do treino) e x de teste (o período para gerar a previsão dos próximos 90 dias).\n",
    "\n",
    "Para treinar o modelo, usaremos todo os dados menos as compras dos ultimos 90 dias, que serão o target do trieno. Para explicitar a tendência mais recente dos dados para o modelo, agrupamos os valores dos últimos 3, 4, 5 e 6 meses criado features (variáveis) com a soma, média e desvio padrão do valor das compras nesses períodos. Afim de transmitir a sazonalidade anual para o modelo, criamos uma feature com a soma das compras dos 3 meses do ano anterior. Essa última melhorou a performance significativamente. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "5220da2b",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-10-17T06:35:32.774563Z",
     "iopub.status.busy": "2021-10-17T06:35:32.773872Z",
     "iopub.status.idle": "2021-10-17T06:35:32.788265Z",
     "shell.execute_reply": "2021-10-17T06:35:32.787172Z",
     "shell.execute_reply.started": "2021-10-17T05:40:30.356329Z"
    },
    "papermill": {
     "duration": 0.037923,
     "end_time": "2021-10-17T06:35:32.788413",
     "exception": false,
     "start_time": "2021-10-17T06:35:32.750490",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "# intervalo de 90 dias\n",
    "delta90 = pd.to_timedelta(90,unit='d')\n",
    "# data mais recente do dataset\n",
    "max_date = df['DT_VENDA'].max()\n",
    "# data de início do target\n",
    "target_start = max_date - delta90\n",
    "# data de início dos dados de treino\n",
    "train_start = target_start - 5*delta90\n",
    "# data de início dos dados de previsão (submissão)\n",
    "sub_start = max_date - 5*delta90"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "dfe81b1f",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-10-17T04:39:03.237208Z",
     "iopub.status.busy": "2021-10-17T04:39:03.236387Z",
     "iopub.status.idle": "2021-10-17T04:39:03.243496Z",
     "shell.execute_reply": "2021-10-17T04:39:03.242352Z",
     "shell.execute_reply.started": "2021-10-17T04:39:03.237155Z"
    },
    "papermill": {
     "duration": 0.020726,
     "end_time": "2021-10-17T06:35:32.830580",
     "exception": false,
     "start_time": "2021-10-17T06:35:32.809854",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Pré-processamento\n",
    "\n",
    "Transformar o dataset original no formato próprio para previsão."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "f247b889",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-10-17T06:35:32.880054Z",
     "iopub.status.busy": "2021-10-17T06:35:32.878081Z",
     "iopub.status.idle": "2021-10-17T06:35:40.641417Z",
     "shell.execute_reply": "2021-10-17T06:35:40.640836Z",
     "shell.execute_reply.started": "2021-10-17T05:46:39.756568Z"
    },
    "papermill": {
     "duration": 7.790031,
     "end_time": "2021-10-17T06:35:40.641554",
     "exception": false,
     "start_time": "2021-10-17T06:35:32.851523",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "x_train_total, y_train_total, x_test = transform_df(df)\n",
    "x_train_split, x_test_split, y_train_split, y_test_split = train_test_split(x_train_total, \n",
    "                                                                            y_train_total, \n",
    "                                                                            random_state=0,\n",
    "                                                                            test_size=0.3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "0822ee66",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-10-17T06:35:40.702890Z",
     "iopub.status.busy": "2021-10-17T06:35:40.702044Z",
     "iopub.status.idle": "2021-10-17T06:35:40.705858Z",
     "shell.execute_reply": "2021-10-17T06:35:40.705387Z",
     "shell.execute_reply.started": "2021-10-17T05:46:59.751755Z"
    },
    "papermill": {
     "duration": 0.043428,
     "end_time": "2021-10-17T06:35:40.705995",
     "exception": false,
     "start_time": "2021-10-17T06:35:40.662567",
     "status": "completed"
    },
    "tags": []
   },
   "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>VALOR_sum</th>\n",
       "      <th>Frequency</th>\n",
       "      <th>valor medio</th>\n",
       "      <th>valor mediana</th>\n",
       "      <th>valor desvio</th>\n",
       "      <th>Recency</th>\n",
       "      <th>valor_90m</th>\n",
       "      <th>valormedio_90m</th>\n",
       "      <th>valor_120m</th>\n",
       "      <th>valormedio_120m</th>\n",
       "      <th>valor_150m</th>\n",
       "      <th>valormedio_150m</th>\n",
       "      <th>valor_180m</th>\n",
       "      <th>valormedio_180m</th>\n",
       "      <th>valorsum_1ano_atras_m</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ID_CLIENTE</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>266.94</td>\n",
       "      <td>1</td>\n",
       "      <td>266.940000</td>\n",
       "      <td>266.940</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>56</td>\n",
       "      <td>266.94</td>\n",
       "      <td>266.94</td>\n",
       "      <td>266.94</td>\n",
       "      <td>266.94</td>\n",
       "      <td>266.94</td>\n",
       "      <td>266.94</td>\n",
       "      <td>266.94</td>\n",
       "      <td>266.94</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>426.45</td>\n",
       "      <td>4</td>\n",
       "      <td>106.612500</td>\n",
       "      <td>105.285</td>\n",
       "      <td>75.738965</td>\n",
       "      <td>151</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>104.69</td>\n",
       "      <td>104.69</td>\n",
       "      <td>121.08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>215.06</td>\n",
       "      <td>3</td>\n",
       "      <td>71.686667</td>\n",
       "      <td>42.790</td>\n",
       "      <td>61.268823</td>\n",
       "      <td>104</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>30.21</td>\n",
       "      <td>30.21</td>\n",
       "      <td>30.21</td>\n",
       "      <td>30.21</td>\n",
       "      <td>30.21</td>\n",
       "      <td>30.21</td>\n",
       "      <td>184.85</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>714.65</td>\n",
       "      <td>7</td>\n",
       "      <td>102.092857</td>\n",
       "      <td>46.440</td>\n",
       "      <td>167.318644</td>\n",
       "      <td>16</td>\n",
       "      <td>532.64</td>\n",
       "      <td>266.32</td>\n",
       "      <td>532.64</td>\n",
       "      <td>266.32</td>\n",
       "      <td>532.64</td>\n",
       "      <td>266.32</td>\n",
       "      <td>532.64</td>\n",
       "      <td>266.32</td>\n",
       "      <td>111.51</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>47.19</td>\n",
       "      <td>1</td>\n",
       "      <td>47.190000</td>\n",
       "      <td>47.190</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>237</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            VALOR_sum  Frequency  valor medio  valor mediana  valor desvio  \\\n",
       "ID_CLIENTE                                                                   \n",
       "0              266.94          1   266.940000        266.940      0.000000   \n",
       "2              426.45          4   106.612500        105.285     75.738965   \n",
       "3              215.06          3    71.686667         42.790     61.268823   \n",
       "4              714.65          7   102.092857         46.440    167.318644   \n",
       "5               47.19          1    47.190000         47.190      0.000000   \n",
       "\n",
       "            Recency  valor_90m  valormedio_90m  valor_120m  valormedio_120m  \\\n",
       "ID_CLIENTE                                                                    \n",
       "0                56     266.94          266.94      266.94           266.94   \n",
       "2               151       0.00            0.00        0.00             0.00   \n",
       "3               104       0.00            0.00       30.21            30.21   \n",
       "4                16     532.64          266.32      532.64           266.32   \n",
       "5               237       0.00            0.00        0.00             0.00   \n",
       "\n",
       "            valor_150m  valormedio_150m  valor_180m  valormedio_180m  \\\n",
       "ID_CLIENTE                                                             \n",
       "0               266.94           266.94      266.94           266.94   \n",
       "2                 0.00             0.00      104.69           104.69   \n",
       "3                30.21            30.21       30.21            30.21   \n",
       "4               532.64           266.32      532.64           266.32   \n",
       "5                 0.00             0.00        0.00             0.00   \n",
       "\n",
       "            valorsum_1ano_atras_m  \n",
       "ID_CLIENTE                         \n",
       "0                            0.00  \n",
       "2                          121.08  \n",
       "3                          184.85  \n",
       "4                          111.51  \n",
       "5                            0.00  "
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train_total.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "a65cdac4",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-10-17T06:35:40.755509Z",
     "iopub.status.busy": "2021-10-17T06:35:40.754858Z",
     "iopub.status.idle": "2021-10-17T06:35:40.759360Z",
     "shell.execute_reply": "2021-10-17T06:35:40.758787Z",
     "shell.execute_reply.started": "2021-10-17T05:43:07.492754Z"
    },
    "papermill": {
     "duration": 0.032024,
     "end_time": "2021-10-17T06:35:40.759502",
     "exception": false,
     "start_time": "2021-10-17T06:35:40.727478",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "ID_CLIENTE\n",
       "0    369.35\n",
       "2    235.81\n",
       "3      0.00\n",
       "4    557.22\n",
       "5      0.00\n",
       "Name: VALOR_TARGET, dtype: float64"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train_total.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "d682ab92",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-10-17T06:35:40.806709Z",
     "iopub.status.busy": "2021-10-17T06:35:40.805880Z",
     "iopub.status.idle": "2021-10-17T06:35:40.822819Z",
     "shell.execute_reply": "2021-10-17T06:35:40.823356Z",
     "shell.execute_reply.started": "2021-10-17T05:42:49.482419Z"
    },
    "papermill": {
     "duration": 0.042086,
     "end_time": "2021-10-17T06:35:40.823526",
     "exception": false,
     "start_time": "2021-10-17T06:35:40.781440",
     "status": "completed"
    },
    "tags": []
   },
   "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>VALOR</th>\n",
       "      <th>Frequency</th>\n",
       "      <th>valor medio</th>\n",
       "      <th>valor mediana</th>\n",
       "      <th>valor desvio</th>\n",
       "      <th>Recency</th>\n",
       "      <th>valor_90m</th>\n",
       "      <th>valormedio_90m</th>\n",
       "      <th>valor_120m</th>\n",
       "      <th>valormedio_120m</th>\n",
       "      <th>valor_150m</th>\n",
       "      <th>valormedio_150m</th>\n",
       "      <th>valor_180m</th>\n",
       "      <th>valormedio_180m</th>\n",
       "      <th>valorsum_1ano_atras_m</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ID_CLIENTE</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>636.29</td>\n",
       "      <td>2</td>\n",
       "      <td>318.145000</td>\n",
       "      <td>318.145</td>\n",
       "      <td>72.414805</td>\n",
       "      <td>84</td>\n",
       "      <td>369.35</td>\n",
       "      <td>369.350</td>\n",
       "      <td>369.35</td>\n",
       "      <td>369.350</td>\n",
       "      <td>636.29</td>\n",
       "      <td>318.145000</td>\n",
       "      <td>636.29</td>\n",
       "      <td>318.145000</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>662.26</td>\n",
       "      <td>6</td>\n",
       "      <td>110.376667</td>\n",
       "      <td>105.285</td>\n",
       "      <td>79.194434</td>\n",
       "      <td>49</td>\n",
       "      <td>235.81</td>\n",
       "      <td>117.905</td>\n",
       "      <td>235.81</td>\n",
       "      <td>117.905</td>\n",
       "      <td>235.81</td>\n",
       "      <td>117.905000</td>\n",
       "      <td>235.81</td>\n",
       "      <td>117.905000</td>\n",
       "      <td>200.68</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>215.06</td>\n",
       "      <td>3</td>\n",
       "      <td>71.686667</td>\n",
       "      <td>42.790</td>\n",
       "      <td>61.268823</td>\n",
       "      <td>195</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1271.87</td>\n",
       "      <td>11</td>\n",
       "      <td>115.624545</td>\n",
       "      <td>59.010</td>\n",
       "      <td>134.009345</td>\n",
       "      <td>16</td>\n",
       "      <td>557.22</td>\n",
       "      <td>139.305</td>\n",
       "      <td>610.01</td>\n",
       "      <td>122.002</td>\n",
       "      <td>1089.86</td>\n",
       "      <td>181.643333</td>\n",
       "      <td>1089.86</td>\n",
       "      <td>181.643333</td>\n",
       "      <td>70.50</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>47.19</td>\n",
       "      <td>1</td>\n",
       "      <td>47.190000</td>\n",
       "      <td>47.190</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>328</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>47.19</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              VALOR  Frequency  valor medio  valor mediana  valor desvio  \\\n",
       "ID_CLIENTE                                                                 \n",
       "0            636.29          2   318.145000        318.145     72.414805   \n",
       "2            662.26          6   110.376667        105.285     79.194434   \n",
       "3            215.06          3    71.686667         42.790     61.268823   \n",
       "4           1271.87         11   115.624545         59.010    134.009345   \n",
       "5             47.19          1    47.190000         47.190      0.000000   \n",
       "\n",
       "            Recency  valor_90m  valormedio_90m  valor_120m  valormedio_120m  \\\n",
       "ID_CLIENTE                                                                    \n",
       "0                84     369.35         369.350      369.35          369.350   \n",
       "2                49     235.81         117.905      235.81          117.905   \n",
       "3               195       0.00           0.000        0.00            0.000   \n",
       "4                16     557.22         139.305      610.01          122.002   \n",
       "5               328       0.00           0.000        0.00            0.000   \n",
       "\n",
       "            valor_150m  valormedio_150m  valor_180m  valormedio_180m  \\\n",
       "ID_CLIENTE                                                             \n",
       "0               636.29       318.145000      636.29       318.145000   \n",
       "2               235.81       117.905000      235.81       117.905000   \n",
       "3                 0.00         0.000000        0.00         0.000000   \n",
       "4              1089.86       181.643333     1089.86       181.643333   \n",
       "5                 0.00         0.000000        0.00         0.000000   \n",
       "\n",
       "            valorsum_1ano_atras_m  \n",
       "ID_CLIENTE                         \n",
       "0                            0.00  \n",
       "2                          200.68  \n",
       "3                            0.00  \n",
       "4                           70.50  \n",
       "5                           47.19  "
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_test.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "36af09a7",
   "metadata": {
    "papermill": {
     "duration": 0.022207,
     "end_time": "2021-10-17T06:35:40.868139",
     "exception": false,
     "start_time": "2021-10-17T06:35:40.845932",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Modelo - ElasticNet\n",
    "\n",
    "Usaremos um modelo linear com regularizações l1 e l2 vizando minimizar o impacto dos ruídos dos dados, do ruído inerente a prever valores futuros a partir de uma quantidade e variedade limitada de dados que muito provavelmente não carrega toda a variância dos dados e o impacto da correlação das variáveis que criamos nos pesos da regressão."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "5bb67f42",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-10-17T06:35:40.916624Z",
     "iopub.status.busy": "2021-10-17T06:35:40.915973Z",
     "iopub.status.idle": "2021-10-17T06:35:42.051727Z",
     "shell.execute_reply": "2021-10-17T06:35:42.053372Z",
     "shell.execute_reply.started": "2021-10-17T05:47:08.488762Z"
    },
    "papermill": {
     "duration": 1.163235,
     "end_time": "2021-10-17T06:35:42.053851",
     "exception": false,
     "start_time": "2021-10-17T06:35:40.890616",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Validação direta RMSE:  242.17955974579021\n"
     ]
    }
   ],
   "source": [
    "en = ElasticNet(random_state=0)\n",
    "en.fit(x_train_split, y_train_split)\n",
    "en_pred = en.predict(x_test_split)\n",
    "# validação direta\n",
    "score_dir = mean_squared_error(y_test_split,en_pred, squared=False)\n",
    "print('Validação direta RMSE: ',score_dir)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "22555b08",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-10-17T06:35:42.192205Z",
     "iopub.status.busy": "2021-10-17T06:35:42.191018Z",
     "iopub.status.idle": "2021-10-17T06:35:50.049505Z",
     "shell.execute_reply": "2021-10-17T06:35:50.050489Z",
     "shell.execute_reply.started": "2021-10-17T05:43:48.50434Z"
    },
    "papermill": {
     "duration": 7.912161,
     "end_time": "2021-10-17T06:35:50.050756",
     "exception": false,
     "start_time": "2021-10-17T06:35:42.138595",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "RMSE:  245.95534755944885 \n",
      "std = : 17.352346333988397\n"
     ]
    }
   ],
   "source": [
    "# validação cruzada\n",
    "score = cross_val_score(en, X=x_train_total, y=y_train_total, cv=5, scoring='neg_root_mean_squared_error')\n",
    "print('RMSE: ',-score.mean(), '\\nstd = :', score.std())"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5ee96eed",
   "metadata": {
    "papermill": {
     "duration": 0.044541,
     "end_time": "2021-10-17T06:35:50.140477",
     "exception": false,
     "start_time": "2021-10-17T06:35:50.095936",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "O desvio padrão alto indica que o modelo está muito sensível ao conjunto de dados usado. Variar a random_seed do train_test_split também altera significativamente o score, o que reforça essa ideia."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "111b2d89",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-10-17T06:35:50.216701Z",
     "iopub.status.busy": "2021-10-17T06:35:50.216054Z",
     "iopub.status.idle": "2021-10-17T06:36:11.071004Z",
     "shell.execute_reply": "2021-10-17T06:36:11.071536Z",
     "shell.execute_reply.started": "2021-10-17T05:44:00.885261Z"
    },
    "papermill": {
     "duration": 20.897408,
     "end_time": "2021-10-17T06:36:11.071856",
     "exception": false,
     "start_time": "2021-10-17T06:35:50.174448",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<module 'matplotlib.pyplot' from '/opt/conda/lib/python3.7/site-packages/matplotlib/pyplot.py'>"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot curva de aprendizado com 70% dos dados\n",
    "title = \"Learning Curves em um split dos dados - ElasticNet\"\n",
    "plot_learning_curve(ElasticNet(random_state=0), title, x_train_split, y_train_split, cv=5, n_jobs=-1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "37bd2732",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-10-17T05:50:18.787013Z",
     "iopub.status.busy": "2021-10-17T05:50:18.786669Z",
     "iopub.status.idle": "2021-10-17T05:50:18.793525Z",
     "shell.execute_reply": "2021-10-17T05:50:18.792199Z",
     "shell.execute_reply.started": "2021-10-17T05:50:18.786978Z"
    },
    "papermill": {
     "duration": 0.024585,
     "end_time": "2021-10-17T06:36:11.121888",
     "exception": false,
     "start_time": "2021-10-17T06:36:11.097303",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Curva de aprendizado com 70% dos dados não indica overfitting, coerente com aprendizado real."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "eeba526b",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-10-17T06:36:11.177457Z",
     "iopub.status.busy": "2021-10-17T06:36:11.176768Z",
     "iopub.status.idle": "2021-10-17T06:36:37.359037Z",
     "shell.execute_reply": "2021-10-17T06:36:37.358528Z",
     "shell.execute_reply.started": "2021-10-17T05:51:03.137921Z"
    },
    "papermill": {
     "duration": 26.212214,
     "end_time": "2021-10-17T06:36:37.359183",
     "exception": false,
     "start_time": "2021-10-17T06:36:11.146969",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<module 'matplotlib.pyplot' from '/opt/conda/lib/python3.7/site-packages/matplotlib/pyplot.py'>"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# curva de aprendizado com todos os dados\n",
    "title = \"Learning Curves nos dados totais - ElasticNet\"\n",
    "plot_learning_curve(ElasticNet(random_state=0), title, x_train_total, y_train_total, cv=5, n_jobs=-1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e4ab5a9d",
   "metadata": {
    "papermill": {
     "duration": 0.026968,
     "end_time": "2021-10-17T06:36:37.473845",
     "exception": false,
     "start_time": "2021-10-17T06:36:37.446877",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Não sei o que ta acontecendo aqui kkkkkkkk"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c3c1f2a9",
   "metadata": {
    "papermill": {
     "duration": 0.028017,
     "end_time": "2021-10-17T06:36:37.529373",
     "exception": false,
     "start_time": "2021-10-17T06:36:37.501356",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Dividindo por canal\n",
    "\n",
    "Os padrões de compras em lojas físicas, por televendas, whatsapp, ecm e ifood são diferentes, por isso, treinamos um modelo com as compras de cada canal e somamos as previsões finais. Essa solução deu uma pontuação muito semelhante (diferença de décimos no leaderboard) a treinar apenas um modelo com os dados de todos os canais, mas como é um pouco mais elaborada, resolvi compartilhar essa."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "83a74cd8",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-10-17T06:36:37.587802Z",
     "iopub.status.busy": "2021-10-17T06:36:37.587091Z",
     "iopub.status.idle": "2021-10-17T06:36:46.051122Z",
     "shell.execute_reply": "2021-10-17T06:36:46.050475Z",
     "shell.execute_reply.started": "2021-10-17T06:02:45.037973Z"
    },
    "papermill": {
     "duration": 8.494717,
     "end_time": "2021-10-17T06:36:46.051278",
     "exception": false,
     "start_time": "2021-10-17T06:36:37.556561",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "canais = df['CANAL'].unique()\n",
    "# define listas para receber os dados separados por canal\n",
    "x_train_aux = [[] for x in range(len(canais))]\n",
    "y_train_aux = [[] for x in range(len(canais))]\n",
    "x_test_aux = [[] for x in range(len(canais))]\n",
    "\n",
    "# divide o dataset por canal\n",
    "for i, canal in enumerate(canais):\n",
    "    df_aux = df[df['CANAL']==canal]\n",
    "    x_train_aux[i], y_train_aux[i], x_test_aux[i] = transform_df(df_aux)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "b3e5aa52",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-10-17T06:36:46.112015Z",
     "iopub.status.busy": "2021-10-17T06:36:46.111024Z",
     "iopub.status.idle": "2021-10-17T06:36:46.114969Z",
     "shell.execute_reply": "2021-10-17T06:36:46.114415Z",
     "shell.execute_reply.started": "2021-10-17T06:03:47.070438Z"
    },
    "papermill": {
     "duration": 0.036322,
     "end_time": "2021-10-17T06:36:46.115113",
     "exception": false,
     "start_time": "2021-10-17T06:36:46.078791",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "FIS (176452, 15)\n",
      "TELEVENDAS (5764, 15)\n",
      "ECM (36846, 15)\n",
      "WHATSAPP (2921, 15)\n",
      "IFOOD (1247, 15)\n"
     ]
    }
   ],
   "source": [
    "for i,canal in zip(range(5),canais): \n",
    "    print(canal, x_train_aux[i].shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "ba9a9af9",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-10-17T06:36:46.176118Z",
     "iopub.status.busy": "2021-10-17T06:36:46.175384Z",
     "iopub.status.idle": "2021-10-17T06:36:46.178199Z",
     "shell.execute_reply": "2021-10-17T06:36:46.177670Z",
     "shell.execute_reply.started": "2021-10-17T06:04:05.794641Z"
    },
    "papermill": {
     "duration": 0.035495,
     "end_time": "2021-10-17T06:36:46.178340",
     "exception": false,
     "start_time": "2021-10-17T06:36:46.142845",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "# cria lista de modelos\n",
    "modelos = [ElasticNet(random_state=0) for x in range(len(canais))]\n",
    "# lista para receber as predições\n",
    "preds = [[] for x in range(len(canais))]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "3dc3eab6",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-10-17T06:36:46.240092Z",
     "iopub.status.busy": "2021-10-17T06:36:46.239451Z",
     "iopub.status.idle": "2021-10-17T06:36:48.910979Z",
     "shell.execute_reply": "2021-10-17T06:36:48.912331Z",
     "shell.execute_reply.started": "2021-10-17T06:04:06.14996Z"
    },
    "papermill": {
     "duration": 2.706531,
     "end_time": "2021-10-17T06:36:48.912614",
     "exception": false,
     "start_time": "2021-10-17T06:36:46.206083",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/lib/python3.7/site-packages/sklearn/linear_model/_coordinate_descent.py:531: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations. Duality gap: 93772119.45816168, tolerance: 87201.48240262679\n",
      "  positive)\n",
      "/opt/conda/lib/python3.7/site-packages/sklearn/linear_model/_coordinate_descent.py:531: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations. Duality gap: 19784006.713751584, tolerance: 48474.09224963906\n",
      "  positive)\n",
      "/opt/conda/lib/python3.7/site-packages/sklearn/linear_model/_coordinate_descent.py:531: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations. Duality gap: 2740330.686439343, tolerance: 6620.535482075572\n",
      "  positive)\n"
     ]
    }
   ],
   "source": [
    "# treina e faz a previsão para cada modelo\n",
    "for i, modelo in enumerate(modelos):\n",
    "    preds[i] = modelo.fit(x_train_aux[i], y_train_aux[i]).predict(x_test_aux[i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "3738045a",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-10-17T06:36:49.034936Z",
     "iopub.status.busy": "2021-10-17T06:36:49.034321Z",
     "iopub.status.idle": "2021-10-17T06:36:49.191609Z",
     "shell.execute_reply": "2021-10-17T06:36:49.192071Z",
     "shell.execute_reply.started": "2021-10-17T06:06:31.330605Z"
    },
    "papermill": {
     "duration": 0.22284,
     "end_time": "2021-10-17T06:36:49.192253",
     "exception": false,
     "start_time": "2021-10-17T06:36:48.969413",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/lib/python3.7/site-packages/pandas/core/frame.py:9139: FutureWarning: Passing 'suffixes' which cause duplicate columns {'0_sum'} in the result is deprecated and will raise a MergeError in a future version.\n",
      "  sort=sort,\n"
     ]
    }
   ],
   "source": [
    "# cria lista para receber as previsões transformadas para dataframe\n",
    "df_preds=[[] for x in range(5)]\n",
    "for i, canal in enumerate(canais):\n",
    "    df_preds[i] = pd.DataFrame(preds[i], index=x_test_aux[i].index)\n",
    "\n",
    "df_sub = pd.read_csv('/kaggle/input/VLabs-DC/sample_submission.csv',header=0)\n",
    "# Join nas previsões dos diferentes canais\n",
    "for prediction in df_preds:\n",
    "     df_sub = df_sub.join(prediction,on='ID_CLIENTE', lsuffix='_sum', rsuffix='_TARGET')\n",
    "\n",
    "# soma as previsões dos canais em uma coluna \"FINAL\"\n",
    "df_sub['FINAL'] = df_sub.iloc[:,1:].sum(axis=1)\n",
    "# drop nas previsões para ficar só a soma\n",
    "df_sub.drop(list(df_sub.columns[1:-1]),axis=1,inplace=True)\n",
    "# corrige o nome das colunas para submissão\n",
    "df_sub.columns = ['ID_CLIENTE','VALOR']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "7d507b23",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-10-17T06:36:49.262370Z",
     "iopub.status.busy": "2021-10-17T06:36:49.261559Z",
     "iopub.status.idle": "2021-10-17T06:36:49.264770Z",
     "shell.execute_reply": "2021-10-17T06:36:49.265174Z",
     "shell.execute_reply.started": "2021-10-17T06:06:32.038225Z"
    },
    "papermill": {
     "duration": 0.044521,
     "end_time": "2021-10-17T06:36:49.265350",
     "exception": false,
     "start_time": "2021-10-17T06:36:49.220829",
     "status": "completed"
    },
    "tags": []
   },
   "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_CLIENTE</th>\n",
       "      <th>VALOR</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>4</td>\n",
       "      <td>414.441001</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>9</td>\n",
       "      <td>91.310892</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>12</td>\n",
       "      <td>297.516923</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>15</td>\n",
       "      <td>223.847468</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>19</td>\n",
       "      <td>87.293143</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>126611</th>\n",
       "      <td>384409</td>\n",
       "      <td>61.378438</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>126612</th>\n",
       "      <td>384411</td>\n",
       "      <td>143.194911</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>126613</th>\n",
       "      <td>384415</td>\n",
       "      <td>59.601366</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>126614</th>\n",
       "      <td>384418</td>\n",
       "      <td>166.557001</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>126615</th>\n",
       "      <td>384420</td>\n",
       "      <td>403.912152</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>126616 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        ID_CLIENTE       VALOR\n",
       "0                4  414.441001\n",
       "1                9   91.310892\n",
       "2               12  297.516923\n",
       "3               15  223.847468\n",
       "4               19   87.293143\n",
       "...            ...         ...\n",
       "126611      384409   61.378438\n",
       "126612      384411  143.194911\n",
       "126613      384415   59.601366\n",
       "126614      384418  166.557001\n",
       "126615      384420  403.912152\n",
       "\n",
       "[126616 rows x 2 columns]"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_sub"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "91957ea9",
   "metadata": {
    "papermill": {
     "duration": 0.02844,
     "end_time": "2021-10-17T06:36:49.322638",
     "exception": false,
     "start_time": "2021-10-17T06:36:49.294198",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Outro meio usado para validar vagamente a predição antes de gastar uma submissão no kaggle foi comparar com a melhor submissão que tínhamos até o momento."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "aa4f0cf3",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-10-17T06:36:49.385004Z",
     "iopub.status.busy": "2021-10-17T06:36:49.384020Z",
     "iopub.status.idle": "2021-10-17T06:36:49.386859Z",
     "shell.execute_reply": "2021-10-17T06:36:49.386345Z"
    },
    "papermill": {
     "duration": 0.035496,
     "end_time": "2021-10-17T06:36:49.387002",
     "exception": false,
     "start_time": "2021-10-17T06:36:49.351506",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "# melhor score na public leaderboard\n",
    "#df_sub5 = pd.read_csv(r'C:\\Users\\mmansur\\VLabs\\Submissions\\best_score.csv')\n",
    "\n",
    "# validação direta\n",
    "#score_dir = mean_squared_error(df_sub5,df_sub, squared=False)\n",
    "#print('Validação direta RMSE: ',score_dir)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ba8d3321",
   "metadata": {
    "papermill": {
     "duration": 0.02856,
     "end_time": "2021-10-17T06:36:49.444481",
     "exception": false,
     "start_time": "2021-10-17T06:36:49.415921",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Esse código não vai rodar aqui, foi feito localmente. Retornou RMSE = 20.26 em relação à melhor previsão que tínhamos (um modelo mais simples, com menos features), o que quer dizer que está no caminho certo, vai melhorar ou piorar um poquinho a melhor solução até o momento. Nesse caso, melhorou o RMSE em ~7 pontos."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8cbfc151",
   "metadata": {
    "papermill": {
     "duration": 0.028449,
     "end_time": "2021-10-17T06:36:49.501823",
     "exception": false,
     "start_time": "2021-10-17T06:36:49.473374",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Submissão"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "da867eac",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-10-17T06:36:49.563124Z",
     "iopub.status.busy": "2021-10-17T06:36:49.562502Z",
     "iopub.status.idle": "2021-10-17T06:36:49.862884Z",
     "shell.execute_reply": "2021-10-17T06:36:49.862339Z",
     "shell.execute_reply.started": "2021-10-17T06:22:24.432685Z"
    },
    "papermill": {
     "duration": 0.331949,
     "end_time": "2021-10-17T06:36:49.863051",
     "exception": false,
     "start_time": "2021-10-17T06:36:49.531102",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "df_sub.to_csv('submission_VLabs.csv', index=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "90dba5a0",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-10-17T06:16:50.277127Z",
     "iopub.status.busy": "2021-10-17T06:16:50.276806Z",
     "iopub.status.idle": "2021-10-17T06:16:50.281318Z",
     "shell.execute_reply": "2021-10-17T06:16:50.280447Z",
     "shell.execute_reply.started": "2021-10-17T06:16:50.277096Z"
    },
    "papermill": {
     "duration": 0.028312,
     "end_time": "2021-10-17T06:36:49.920197",
     "exception": false,
     "start_time": "2021-10-17T06:36:49.891885",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Próximos passos\n",
    "\n",
    "Pontos a melhorar:\n",
    "- investigar cada feature individualmente, se está ajudando ou atrapalhando. \n",
    "- suavizar outliers vizando melhorar a performance de modelo linear\n",
    "- testamos uma normalização que não funcionou bem, há mais para explorar em relação a normalizações\n",
    "- investigar distribuição das variáveis. Aplicar transformações vizando aproximar de uma distriução normal pode melhorar modelos lineares\n",
    "- hyperparameter tuning, principalmente alpha e l1_rate\n",
    "- produzir diferentes modelos e stackar utilizando as soluções como input"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.10"
  },
  "papermill": {
   "default_parameters": {},
   "duration": 89.722183,
   "end_time": "2021-10-17T06:36:50.660078",
   "environment_variables": {},
   "exception": null,
   "input_path": "__notebook__.ipynb",
   "output_path": "__notebook__.ipynb",
   "parameters": {},
   "start_time": "2021-10-17T06:35:20.937895",
   "version": "2.3.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}