competitions/Predict-future-sales.ipynb

{"cells":[{"metadata":{},"cell_type":"markdown","source":"# Vendas Futuras\n\nO notebook a seguir descreve os estudos e os passos levados para realizar a competição [Predict Future Sales no kaggle](https://www.kaggle.com/c/competitive-data-science-predict-future-sales/overview) que procura predizer as futuras vendas de uma rede de lojas. O notebook a seguir explica os passos levados para uma aplicação dos regressores Random Forest e XGBoost na análise do problema.\n\nEste notebook foi construindo utilizando diversas referências que podem ser encontradas nos links a seguir:\n\n1. https://machinelearningmastery.com/basic-feature-engineering-time-series-data-python/\n2. https://www.kaggle.com/gemartin/load-data-reduce-memory-usage\n3. https://mlwhiz.com/blog/2017/12/26/win_a_data_science_competition/\n4. https://www.kaggle.com/alessandrosolbiati/using-xgboost-for-time-series-prediction-top-20\n5. https://www.kaggle.com/josemarrugo/predictionsales"},{"metadata":{},"cell_type":"markdown","source":"## Análise dos dados\n\nAntes de tudo, é interessante analisarmos com o que estamos trabalhando, a base e o objetivo dela. A base é composta de vendas realizadas ao longo de 34 meses, cada elemento da base representa uma venda, com informações como identificador do item, identificador da loja onde foi vendido, preço, quantidade de produtos, a categoria do produto, entre outras informações. O propósito desta competição é, a partir dos dados ao longo dos 34 meses, prever vendas do 35º mês. Com as informações disponíveis é possível tentar entender um possível padrão de vendas que generalize bem o cenário e usá-lo para prever as vendas futuras. Os dados da competição são distribuídos em 6 partes:\n\n* **shops**: contém o nome de cada loja e seu id;\n\n* **items**: contém o nome de cada item, seu id, e a qual categoria ele pertence;\n\n* **item_categories**: contém as categorias dos itens;\n\n* **sales_train**: contém os dados das vendas, possui o id do item, o id da loja na qual a venda foi realizada, o dia e o mês da venda, o preço do item e a quantidade de itens vendidos;\n\n* **sample_submission**: um exemplo de submissão;\n\n* **test**: contém o id da loja e do item no qual se deseja prever a venda no 35º mês.\n\nAlgumas informações não fazem muita diferença na análise, como o nome do produto e da loja, já que eles possuem id's que servem para identificá-los."},{"metadata":{},"cell_type":"markdown","source":"Começamos importando algumas bibliotecas utilizadas na análise, como a [pandas](https://pandas.pydata.org/docs/reference/index.html), o [numpy](https://numpy.org/doc/stable/reference/index.html) e o [matplotlib](https://matplotlib.org/). Depois, abrimos os arquivos com os dados das vendas."},{"metadata":{"trusted":true},"cell_type":"code","source":"import pandas as pd\nimport numpy as np\nimport os\nimport matplotlib.pyplot as plt","execution_count":1,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def open_csv(file):\n    path = '/kaggle/input/competitive-data-science-predict-future-sales'\n    return pd.read_csv(os.path.join(path, file))\n\n# Abre os arquivos\nitems = open_csv('items.csv')\nshops = open_csv('shops.csv')\nitem_categories = open_csv('item_categories.csv')\ntrain = open_csv('sales_train.csv')\nsubmission = open_csv('sample_submission.csv')\ntest = open_csv('test.csv')","execution_count":2,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"Abaixo podemos visualizar a quantidade de registros e o tamanho dos dataframes:"},{"metadata":{"trusted":true},"cell_type":"code","source":"print(f'Número de elementos no dataframe de treino: {train.shape[0]}')\nprint(f'Quantidade de lojas diferentes: {shops.shape[0]}')\nprint(f'Quantidade de itens diferentes: {items.shape[0]}')\nprint(f'Quantidade de categorias diferentes: {item_categories.shape[0]}')","execution_count":3,"outputs":[{"output_type":"stream","text":"Número de elementos no dataframe de treino: 2935849\nQuantidade de lojas diferentes: 60\nQuantidade de itens diferentes: 22170\nQuantidade de categorias diferentes: 84\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"print(train.columns)","execution_count":4,"outputs":[{"output_type":"stream","text":"Index(['date', 'date_block_num', 'shop_id', 'item_id', 'item_price',\n       'item_cnt_day'],\n      dtype='object')\n","name":"stdout"}]},{"metadata":{},"cell_type":"markdown","source":"O dataframe possui 6 colunas, que são descritas abaixo:\n* **date**: data da venda no formato dd/mm/aa;\n* **date_block_num**: mês da venda representado em uma variável de 0 até 33;\n* **shop_id**: identificador da loja;\n* **item_id**: identificador do item;\n* **item_price**: preço do item;\n* **item_cnt_day**: quantidade de itens vendidos."},{"metadata":{},"cell_type":"markdown","source":"## Pré-processamento\n\nNos próximos blocos, são realizados alguns tratamentos iniciais no dataset. Tiramos features que foram duplicadas na junção de arquivos, dados desnecessários, como nome dos itens. Como o propósito é a predição de um mês inteiro de vendas, os dados serão agrupados por mês utilizando a informação contida na variável **date_block_num**, logo, o dia da venda também não é útil e será retirado do dataframe."},{"metadata":{"trusted":true},"cell_type":"code","source":"# Remove dados duplicados\nattr = ['date', 'date_block_num', 'shop_id', 'item_id', 'item_cnt_day']\nold_size = train.shape[0]\ntrain.drop_duplicates(attr, inplace=True)\nprint(f'Quantidade de elementos duplicados: {old_size - train.shape[0]}')","execution_count":5,"outputs":[{"output_type":"stream","text":"Quantidade de elementos duplicados: 24\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"# Retira dados sem relevância para a análise\ntrain.drop(columns=['date'], inplace=True)\nitems.drop(columns=['item_name'], inplace=True)\ntest.drop(columns=['ID'], inplace=True)","execution_count":6,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"Foram retirados alguns elementos **outliers** do dataframe, que são valores atípicos ou prováveis erros na coleta dos dados. Os valores retirados foram em relação ao preço dos itens, e a quantidade de itens retirados. A partir da análise feita abaixo, foi identificada a existência de valores negativos, que não fazem sentido para essas variáveis, e de valores muito distantes das médias, que foram selecionados a partir de um limite definido olhando para os gráficos."},{"metadata":{"trusted":true},"cell_type":"code","source":"# Retira outliers\nold_size = train.shape[0]\nprint('Vendas:')\nprint(train['item_cnt_day'].describe())\ntrain.boxplot(column='item_cnt_day')\nplt.show()\nprint('\\nPreço:')\nprint(train['item_price'].describe())\ntrain.boxplot(column='item_price')\nplt.show()","execution_count":7,"outputs":[{"output_type":"stream","text":"Vendas:\ncount    2.935825e+06\nmean     1.242643e+00\nstd      2.618845e+00\nmin     -2.200000e+01\n25%      1.000000e+00\n50%      1.000000e+00\n75%      1.000000e+00\nmax      2.169000e+03\nName: item_cnt_day, dtype: float64\n","name":"stdout"},{"output_type":"display_data","data":{"text/plain":"<Figure size 432x288 with 1 Axes>","image/png":"iVBORw0KGgoAAAANSUhEUgAAAX0AAAD5CAYAAADLL+UrAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+j8jraAAATnElEQVR4nO3df2zc933f8eebpCIbUmXIscPZllIZhbVSZFZ3Yb3AdQuygiunWWN3WwZzRq3AxFQkMaHWgV1nBNYBxW1GgjhwksatEtpWMJeGsy2L0Dp1NJlsqzaNI3VuJJuxrc2OzVqwVihrRXmRQ+q9P/iVcJIokUdKPNGf5wM43Pfe9/3xPuH04pef7+eOkZlIksrQ0uwGJEmLx9CXpIIY+pJUEENfkgpi6EtSQdqa3cBsrrjiily3bl2z25DOcPToUVasWNHsNqQZ7d279+8y88rT6xd96K9bt449e/Y0uw3pDKOjo/T09DS7DWlGEfGDmeoO70hSQQx9SSqIoS9JBTH0Jakghr4kFcTQlxo0PDxMV1cXGzdupKuri+Hh4Wa3JM3ZRT9lU7qYDA8PMzg4yNDQEFNTU7S2ttLf3w9AX19fk7uTZueZvtSAWq3G0NAQvb29tLW10dvby9DQELVardmtSXNi6EsNGBsb46abbjqldtNNNzE2NtakjqTGGPpSAzo6Oti9e/cptd27d9PR0dGkjqTGGPpSAwYHB+nv72dkZITJyUlGRkbo7+9ncHCw2a1Jc+KFXKkBJy7WDgwMMDY2RkdHB7VazYu4WjLiYv8bud3d3ekXruli5Beu6WIWEXszs/v0usM7klQQQ1+SCmLoS1JBDH1JKoihL0kFMfQlqSCGviQVZNbQj4i1ETESEWMR8XxEbK3ql0fEzoh4ubpfXbfNpyLiQES8GBGb6urvj4h91XOfj4i4MC9LkjSTuZzpTwKfzMwO4APAJyJiA3A/sCszrwN2VY+pnrsd6ARuAb4UEa3Vvh4GtgDXVbdbzuNrkSTNYtbQz8yDmfnX1fIRYAy4BrgV2F6tth24rVq+FXgiM49l5ivAAeCGiLgKWJWZ387pjwF/tW4bSdIiaGhMPyLWAT8LfAdoz8yDMP2DAXhPtdo1wOt1m41XtWuq5dPrkqRFMucvXIuIlcB/BX4zM//hHMPxMz2R56jPdKwtTA8D0d7ezujo6FzblBbNxMSE700tOXMK/YhYxnTgP56Z/60qvxkRV2XmwWro5lBVHwfW1m2+Bnijqq+ZoX6GzNwGbIPpL1zzS610MfIL17QUzWX2TgBDwFhmPlj31A5gc7W8GfhGXf32iFgeEdcyfcH22WoI6EhEfKDa551120iSFsFczvR/Hvh1YF9EPFfV/h3wAPBkRPQDrwEfAcjM5yPiSeAFpmf+fCIzp6rtPgY8BlwKfLO6SZIWyayhn5m7mXk8HmDjWbapAWf8pejM3AN0NdKgJOn88RO5klQQQ1+SCmLoS1JBDH1JKoihL0kFMfQlqSCGviQVxNCXpIIY+pJUEENfkgpi6EtSQQx9SSqIoS9JBTH0Jakghr4kFcTQl6SCGPqSVBBDX5IKYuhLUkEMfUkqiKEvSQUx9CWpIIa+JBXE0Jekghj6klQQQ1+SCmLoS1JBDH1JKoihL0kFMfQlqSCGviQVxNCXpIIY+pJUEENfkgpi6EtSQQx9SSrIrKEfEY9ExKGI2F9X+w8R8bcR8Vx1+5W65z4VEQci4sWI2FRXf39E7Kue+3xExPl/OZKkc5nLmf5jwC0z1D+XmddXt6cAImIDcDvQWW3zpYhordZ/GNgCXFfdZtqnJOkCmjX0M/PPgMNz3N+twBOZeSwzXwEOADdExFXAqsz8dmYm8FXgtvk2LUman4WM6d8dEd+rhn9WV7VrgNfr1hmvatdUy6fXJUmLqG2e2z0M/C6Q1f1ngbuAmcbp8xz1GUXEFqaHgmhvb2d0dHSebUoXzsTEhO9NLTnzCv3MfPPEckR8Gfij6uE4sLZu1TXAG1V9zQz1s+1/G7ANoLu7O3t6eubTpnRBjY6O4ntTS828hneqMfoTfg04MbNnB3B7RCyPiGuZvmD7bGYeBI5ExAeqWTt3At9YQN+SpHmY9Uw/IoaBHuCKiBgHfgfoiYjrmR6ieRX4DYDMfD4ingReACaBT2TmVLWrjzE9E+hS4JvVTZK0iGYN/czsm6E8dI71a0BthvoeoKuh7iRJ55WfyJWkghj6klQQQ1+SCmLoS1JBDH1JKoihL0kFMfQlqSCGviQVxNCXpIIY+pJUEENfkgpi6EtSQQx9SSqIoS9JBTH0Jakghr4kFcTQl6SCGPqSVBBDX5IKYuhLUkEMfUkqiKEvSQUx9CWpIIa+JBXE0Jekghj6klQQQ1+SCmLoS1JBDH1JKoihL0kFMfQlqSCGviQVxNCXpIIY+pJUEENfkgpi6EtSQQx9SSrIrKEfEY9ExKGI2F9XuzwidkbEy9X96rrnPhURByLixYjYVFd/f0Tsq577fETE+X850oU3PDxMV1cXGzdupKuri+Hh4Wa3JM1Z2xzWeQz4IvDVutr9wK7MfCAi7q8e/3ZEbABuBzqBq4H/ERHrM3MKeBjYAvwV8BRwC/DN8/VCpMUwPDzM4OAgQ0NDTE1N0draSn9/PwB9fX1N7k6a3axn+pn5Z8Dh08q3Atur5e3AbXX1JzLzWGa+AhwAboiIq4BVmfntzEymf4DchrTE1Go1hoaG6O3tpa2tjd7eXoaGhqjVas1uTZqTuZzpz6Q9Mw8CZObBiHhPVb+G6TP5E8ar2o+r5dPrM4qILUz/VkB7ezujo6PzbFM6v8bGxpiammJ0dJSJiQlGR0eZmppibGzM96mWhPmG/tnMNE6f56jPKDO3AdsAuru7s6en57w0Jy1UR0cHra2t9PT0MDo6Sk9PDyMjI3R0dOD7VEvBfGfvvFkN2VDdH6rq48DauvXWAG9U9TUz1KUlZXBwkP7+fkZGRpicnGRkZIT+/n4GBweb3Zo0J/M9098BbAYeqO6/UVf/w4h4kOkLudcBz2bmVEQciYgPAN8B7gS+sKDOpSY4cbF2YGCAsbExOjo6qNVqXsTVkhHT11XPsULEMNADXAG8CfwO8N+BJ4H3Aq8BH8nMw9X6g8BdwCTwm5n5zarezfRMoEuZnrUzkLMdnOnhnT179szjpUkX1onhHeliFBF7M7P79PqsZ/qZebZTmI1nWb8GnDGVITP3AF2zHU+SdOH4iVxJKoihL0kFMfQlqSCGviQVxNCXpIIY+pJUEENfkgpi6EtSQQx9SSqIoS9JBTH0Jakghr4kFcTQl6SCGPqSVBBDX5IKYuhLUkEMfUkqiKEvSQUx9CWpIIa+JBXE0Jekghj6UoOGh4fp6upi48aNdHV1MTw83OyWpDlra3YD0lIyPDzM4OAgQ0NDTE1N0draSn9/PwB9fX1N7k6anWf6UgNqtRpDQ0P09vbS1tZGb28vQ0ND1Gq1ZrcmzYmhLzVgbGyM8fHxU4Z3xsfHGRsba3Zr0pwY+lIDrr76agYGBjh69CiZydGjRxkYGODqq69udmvSnBj6UgPeeustJiYmGBgY4KmnnmJgYICJiQneeuutZrcmzYmhLzXg8OHD3HfffTzyyCN86EMf4pFHHuG+++7j8OHDzW5NmhNDX2pQb28v+/fvZ9euXezfv5/e3t5mtyTNmaEvNWDNmjVs3ryZkZERJicnGRkZYfPmzaxZs6bZrUlz4jx9qQGf/vSn2bp1K3fddRevvfYa733ve5mcnOSzn/1ss1uT5sQzfakBfX19PPTQQ6xYsQKAFStW8NBDD/nBLC0Zhr4kFcTQlxowPDzM1q1bOXr0KABHjx5l69atfv+OlozIzGb3cE7d3d25Z8+eZrchAbB27VqmpqZ4/PHHT373zh133EFrayuvv/56s9uTToqIvZnZfXrdC7lSA8bHx/nwhz/MBz/4QY4dO8by5cvZtGkTO3bsaHZr0pwsaHgnIl6NiH0R8VxE7Klql0fEzoh4ubpfXbf+pyLiQES8GBGbFtq81Aw7duzg2LFjABw7dszA15JyPsb0ezPz+rpfI+4HdmXmdcCu6jERsQG4HegEbgG+FBGt5+H40qLr7OxkeHiYzs7OZrciNeRCXMi9FdheLW8HbqurP5GZxzLzFeAAcMMFOL50QbW0tPDSSy/R19fHSy+9REuL8yG0dCx0TD+Bb0VEAn+QmduA9sw8CJCZByPiPdW61wB/VbfteFU7Q0RsAbYAtLe3Mzo6usA2pfOnra2Nyy+/nEOHDvHud7+bw4cP8/bbb/s+1ZKw0ND/+cx8owr2nRHx/XOsGzPUZpw6VP3w2AbTs3d6enoW2KZ0/rz99tvce++9bNiwgRdeeIFPfvKTAPg+1VKwoNDPzDeq+0MR8XWmh2vejIirqrP8q4BD1erjwNq6zdcAbyzk+FKznAh6aamZ92BkRKyIiJ84sQz8MrAf2AFsrlbbDHyjWt4B3B4RyyPiWuA64Nn5Hl9qhs7OTq688spTaldeeaUXdLVkLOQKVDuwOyL+hunw/uPM/BPgAeDmiHgZuLl6TGY+DzwJvAD8CfCJzJxaSPPSYhscHGTlypU888wz7Ny5k2eeeYaVK1cyODjY7NakOfETuVKDNm3axM6dO8lMIoKbb76Zp59+utltSac42ydynWsmNWBgYICdO3eenKbZ0tLCzp07GRgYaHJn0tx4pi81oK2tjczkM5/5zMnZO/feey8RweTkZLPbk07yTF86D6ampqjVatxzzz1ccskl3HPPPdRqNaamvDylpcHQl6SC+C2bUgNaWloYHBzkXe96Fxs2bODBBx9kcHDQr2LQkmHoSw34+Mc/zhe/+MUzPpx19913N6kjqTGenkgNuPHGG1m1ahXLli0DYNmyZaxatYobb7yxyZ1Jc2PoSw2o1WoMDAywfv16WlpaWL9+PQMDA9RqtWa3Js2JUzalBrS0tLBy5Up+9KMf8eMf/5hly5ZxySWXMDExwfHjx5vdnnSSUzal86ClpYUjR46cDPjjx49z5MgRL+RqyfCdKjXgxHz8VatWnXLvPH0tFYa+1KDly5dz2WWX0dLSwmWXXcby5cub3ZI0Z4a+NE8X+/UwaSbO05cadOzYMV599VWAk/fSUuGZviQVxNCXpIIY+pJUEENfmofVq1fT0tLC6tWrm92K1BAv5Erz8MMf/vCUe2mp8Exfkgpi6EtSQQx9SSqIoS9JBTH0Jakghr4kFcTQl6SCGPqSVBBDX5IKYuhLUkEMfUkqiKEvSQUx9CWpIIa+JBXE0Jekghj6klQQQ1+SCrLooR8Rt0TEixFxICLuX+zjS1LJFjX0I6IV+D3gg8AGoC8iNixmD5JUssU+078BOJCZ/zsz3waeAG5d5B4kqViL/YfRrwFer3s8Dvyz01eKiC3AFoD29nZGR0cXpTm9cwz8YOCC7Lfrsa6zPve+7e+7IMf8wk9+4YLsV2Va7NCPGWp5RiFzG7ANoLu7O3t6ei5wW3qn2ce+C7LfiJnewtMyz3grSxedxR7eGQfW1j1eA7yxyD1IUrEWO/S/C1wXEddGxLuA24Edi9yDNG/Lly9vqC5dbBY19DNzErgbeBoYA57MzOcXswdpIR599FGWLVt2Sm3ZsmU8+uijTepIasyiz9PPzKcyc31m/lRm1hb7+NJC9PX1sX37djo7O2lpaaGzs5Pt27fT19fX7NakOYmL/eJTd3d37tmzp9ltSGcYHR3FSQa6WEXE3szsPr3u1zBIUkEMfUkqiKEvSQUx9CWpIIa+JBXkop+9ExH/B/hBs/uQZnAF8HfNbkI6i5/MzCtPL170oS9drCJiz0xT4qSLmcM7klQQQ1+SCmLoS/O3rdkNSI1yTF+SCuKZviQVxNCXpIIY+pJUEENfS1ZE/GV1vy4i/k2z+6kXER+NiKsbWL8nIv7oQvYkgaGvJSwzb6wW1wEXVegDHwXmHPrSYjH0tWRFxES1+ADwCxHxXET8VkS0RsRnIuK7EfG9iPiNav2eiPjTiHgyIl6KiAci4o6IeDYi9kXET53jWO0R8fWI+JvqdmP1G8ZYRHw5Ip6PiG9FxKUR8a+AbuDxqqdLz7LPWyLi+xGxG/gXdfUbIuIvI+J/Vvf/uKr/eURcX7feX0TEP1nwP6SKYujrneB+4M8z8/rM/BzQD/x9Zv4c8HPAv42Ia6t1fwbYCrwP+HVgfWbeAHwFGDjHMT4P/Glm/gzwT4ETf9v5OuD3MrMT+L/Av8zM/wLsAe6oevp/p+8sIi4Bvgz8KvALwD+qe/r7wC9m5s8C/x74j1X9K0z/BkFErAeWZ+b35vIPJJ1g6Oud6JeBOyPiOeA7wLuZDmeA72bmwcw8Bvwv4FtVfR/Tw0Rn80vAwwCZOZWZf1/VX8nM56rlvbPso95PV9u+nNMflvnPdc9dBnwtIvYDnwM6q/rXgH8eEcuAu4DH5ngs6aS2ZjcgXQABDGTm06cUI3qAY3Wl43WPjzO//w/1+5sCZhzKOYuzfTLyd4GRzPy1iFgHjAJk5lsRsRO4FfjXTA8hSQ3xTF/vBEeAn6h7/DTwseqMmIhYHxErFniMXcDHqv21RsSqBns63feBa+uuI/TVPXcZ8LfV8kdP2+4rTA81fTczD8+hb+kUhr7eCb4HTFYXWH+L6WB8AfjraojkD1j4b7Vbgd6I2Mf0ME7nLOs/Bvz+2S7kZuaPgC3AH1cXcuv/ZsSngf8UEX8BtJ623V7gH4BH5/tCVDa/e0daQqq5/6PAT2fm8Sa3oyXIM31piYiIO5m+MD1o4Gu+PNOX6kTEIPCR08pfy8zaAvb5deDa08q/ffqFZmkxGPqSVBCHdySpIIa+JBXE0Jekghj6klSQ/w/dS2Yf4mFifgAAAABJRU5ErkJggg==\n"},"metadata":{"needs_background":"light"}},{"output_type":"stream","text":"\nPreço:\ncount    2.935825e+06\nmean     8.908558e+02\nstd      1.729806e+03\nmin     -1.000000e+00\n25%      2.490000e+02\n50%      3.990000e+02\n75%      9.990000e+02\nmax      3.079800e+05\nName: item_price, dtype: float64\n","name":"stdout"},{"output_type":"display_data","data":{"text/plain":"<Figure size 432x288 with 1 Axes>","image/png":"iVBORw0KGgoAAAANSUhEUgAAAYoAAAD5CAYAAAA5v3LLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+j8jraAAAWuUlEQVR4nO3df4yd1X3n8fcnNssiSFh+JCNik5otRGuw2kTMOkRhteN1ZWgVCboCYXZVXMWSU0RQsmJFof6DNshS0DZlQ7eh63QQBnUNLm0EG8E6Lnja9S4BTAQY47J4QwIDFiQySzFa2Nh89497Jr0exs+Mf8144P2Srua53+ecc8+17ugzzznPeFJVSJJ0IB+Z6QlIko5tBoUkqZNBIUnqZFBIkjoZFJKkTnNnegJH2umnn14LFiyY6WlIE3r77bc58cQTZ3oa0vs8+eSTP6uqj0907gMXFAsWLGDr1q0zPQ1pQiMjIwwNDc30NKT3SfKTA51z6UmS1MmgkCR1MigkSZ0MCklSJ4NCktTJoJCmwfr161m0aBFLly5l0aJFrF+/fqanJE3ZB+72WOlYs379elavXs3w8DD79u1jzpw5rFy5EoArr7xyhmcnTc4rCukoW7NmDcPDwyxZsoS5c+eyZMkShoeHWbNmzUxPTZoSg0I6ynbs2MGFF164X+3CCy9kx44dMzQj6eAYFNJRtnDhQrZs2bJfbcuWLSxcuHCGZiQdnEmDIsk/TvJ4kqeTbE/yB61+apJNSV5oX0/p63Njkp1Jnk9yUV/9/CTb2rnbkqTVj09yb6s/lmRBX58V7TVeSLLiSL55aTqsXr2alStXsnnzZvbu3cvmzZtZuXIlq1evnumpSVMylc3sd4F/VVV7khwHbEnyEPCvgYer6htJbgBuAH43ybnAcuA84JPAXyf5dFXtA24HVgE/AB4ELgYeAlYCb1TV2UmWA7cAVyQ5FbgJGAQKeDLJA1X1xhH7F5COsrEN62uvvZYdO3awcOFC1qxZ40a2Zo1JryiqZ097elx7FHAJsK7V1wGXtuNLgHuq6t2qehHYCSxOcgbwsap6tHp/qPuucX3GxroPWNquNi4CNlXV7hYOm+iFizSrXHnllTz77LM8/PDDPPvss4aEZpUp3R6bZA7wJHA28CdV9ViSgaraBVBVu5J8ojWfR++KYcxoq/28HY+vj/V5uY21N8mbwGn99Qn69M9vFb0rFQYGBhgZGZnK25Km3Z49e/x8ataZUlC0ZaPPJPknwHeTLOponomG6Kgfap/++a0F1gIMDg6W/42zjlX+N+OajQ7qrqeq+j/ACL3ln9fachLt6+ut2ShwZl+3+cCrrT5/gvp+fZLMBU4GdneMJUmaJlO56+nj7UqCJCcAvwb8HfAAMHYX0grg/nb8ALC83cl0FnAO8HhbpnoryQVt/+GqcX3GxroMeKTtY2wEliU5pd1VtazVJEnTZCpLT2cA69o+xUeADVX1vSSPAhuSrAReAi4HqKrtSTYAzwF7gWva0hXA1cCdwAn07nZ6qNWHgbuT7KR3JbG8jbU7yc3AE63d16tq9+G8YUnSwUnvB/cPjsHBwfJPoepY5R6FjlVJnqyqwYnO+ZvZkqROBoUkqZNBIUnqZFBIkjoZFJKkTgaFJKmTQSFJ6mRQSJI6GRSSpE4GhSSpk0EhSepkUEiSOhkUkqROBoUkqZNBIUnqZFBIkjoZFJKkTgaFJKmTQSFJ6mRQSJI6GRSSpE4GhSSpk0EhSeo0aVAkOTPJ5iQ7kmxP8tVW//0kryR5qj1+o6/PjUl2Jnk+yUV99fOTbGvnbkuSVj8+yb2t/liSBX19ViR5oT1WHMk3L0ma3NwptNkLXFdVP0zyUeDJJJvauVur6g/7Gyc5F1gOnAd8EvjrJJ+uqn3A7cAq4AfAg8DFwEPASuCNqjo7yXLgFuCKJKcCNwGDQLXXfqCq3ji8ty1JmqpJryiqaldV/bAdvwXsAOZ1dLkEuKeq3q2qF4GdwOIkZwAfq6pHq6qAu4BL+/qsa8f3AUvb1cZFwKaq2t3CYRO9cJEkTZOD2qNoS0KfBR5rpa8keSbJHUlOabV5wMt93UZbbV47Hl/fr09V7QXeBE7rGEuSNE2msvQEQJKTgL8EvlZVf5/kduBmektCNwPfBL4EZILu1VHnEPv0z20VvSUtBgYGGBkZ6Xwv0kzZs2ePn0/NOlMKiiTH0QuJP6+qvwKoqtf6zn8H+F57Ogqc2dd9PvBqq8+foN7fZzTJXOBkYHerD43rMzJ+flW1FlgLMDg4WENDQ+ObSMeEkZER/HxqtpnKXU8BhoEdVfVHffUz+pr9JvBsO34AWN7uZDoLOAd4vKp2AW8luaCNeRVwf1+fsTuaLgMeafsYG4FlSU5pS1vLWk2SNE2mckXxBeC3gG1Jnmq13wOuTPIZektBPwa+DFBV25NsAJ6jd8fUNe2OJ4CrgTuBE+jd7fRQqw8DdyfZSe9KYnkba3eSm4EnWruvV9XuQ3urkqRDMWlQVNUWJt4reLCjzxpgzQT1rcCiCervAJcfYKw7gDsmm6ck6ejwN7MlSZ0MCklSJ4NCktTJoJAkdTIoJEmdDApJUieDQpLUyaCQJHUyKCRJnQwKSVIng0KS1MmgkCR1MigkSZ0MCklSJ4NCktTJoJAkdTIoJEmdDApJUieDQpLUyaCQJHUyKCRJnQwKSVIng0KS1GnSoEhyZpLNSXYk2Z7kq61+apJNSV5oX0/p63Njkp1Jnk9yUV/9/CTb2rnbkqTVj09yb6s/lmRBX58V7TVeSLLiSL55SdLkpnJFsRe4rqoWAhcA1yQ5F7gBeLiqzgEebs9p55YD5wEXA99OMqeNdTuwCjinPS5u9ZXAG1V1NnArcEsb61TgJuBzwGLgpv5AkiQdfZMGRVXtqqoftuO3gB3APOASYF1rtg64tB1fAtxTVe9W1YvATmBxkjOAj1XVo1VVwF3j+oyNdR+wtF1tXARsqqrdVfUGsIl/CBdJ0jQ4qD2KtiT0WeAxYKCqdkEvTIBPtGbzgJf7uo222rx2PL6+X5+q2gu8CZzWMZYkaZrMnWrDJCcBfwl8rar+vm0vTNh0glp11A+1T//cVtFb0mJgYICRkZEDzU2aUXv27PHzqVlnSkGR5Dh6IfHnVfVXrfxakjOqaldbVnq91UeBM/u6zwdebfX5E9T7+4wmmQucDOxu9aFxfUbGz6+q1gJrAQYHB2toaGh8E+mYMDIygp9PzTZTuespwDCwo6r+qO/UA8DYXUgrgPv76svbnUxn0du0frwtT72V5II25lXj+oyNdRnwSNvH2AgsS3JK28Re1mqSpGkylSuKLwC/BWxL8lSr/R7wDWBDkpXAS8DlAFW1PckG4Dl6d0xdU1X7Wr+rgTuBE4CH2gN6QXR3kp30riSWt7F2J7kZeKK1+3pV7T7E9ypJOgSTBkVVbWHivQKApQfoswZYM0F9K7Bogvo7tKCZ4NwdwB2TzVOSdHT4m9mSpE4GhSSpk0EhSepkUEiSOhkUkqROBoUkqZNBIUnqZFBIkjoZFJKkTgaFJKmTQSFJ6mRQSJI6GRSSpE4GhSSpk0EhSepkUEiSOhkUkqROBoUkqZNBIUnqZFBIkjoZFJKkTgaFJKmTQSFJ6jRpUCS5I8nrSZ7tq/1+kleSPNUev9F37sYkO5M8n+Sivvr5Sba1c7clSasfn+TeVn8syYK+PiuSvNAeK47Um5YkTd1UrijuBC6eoH5rVX2mPR4ESHIusBw4r/X5dpI5rf3twCrgnPYYG3Ml8EZVnQ3cCtzSxjoVuAn4HLAYuCnJKQf9DiVJh2XSoKiqvwV2T3G8S4B7qurdqnoR2AksTnIG8LGqerSqCrgLuLSvz7p2fB+wtF1tXARsqqrdVfUGsImJA0uSdBQdzh7FV5I805amxn7Snwe83NdmtNXmtePx9f36VNVe4E3gtI6xJEnTaO4h9rsduBmo9vWbwJeATNC2OuocYp/9JFlFb1mLgYEBRkZGOqYuzZw9e/b4+dSsc0hBUVWvjR0n+Q7wvfZ0FDizr+l84NVWnz9Bvb/PaJK5wMn0lrpGgaFxfUYOMJ+1wFqAwcHBGhoamqiZNONGRkbw86nZ5pCWntqew5jfBMbuiHoAWN7uZDqL3qb141W1C3gryQVt/+Eq4P6+PmN3NF0GPNL2MTYCy5Kc0pa2lrWaJGkaTXpFkWQ9vZ/sT08ySu9OpKEkn6G3FPRj4MsAVbU9yQbgOWAvcE1V7WtDXU3vDqoTgIfaA2AYuDvJTnpXEsvbWLuT3Aw80dp9vaqmuqkuSTpC0vvh/YNjcHCwtm7dOtPTkCbk0pOOVUmerKrBic75m9mSpE4GhSSpk0EhSepkUEiSOhkUkqROBoUkqZNBIUnqZFBIkjoZFJKkTgaFJKmTQSFJ6mRQSJI6GRSSpE4GhSSpk0EhSepkUEiSOhkUkqROBoUkqZNBIUnqZFBIkjoZFJKkTgaFJKmTQSFJ6mRQSJI6TRoUSe5I8nqSZ/tqpybZlOSF9vWUvnM3JtmZ5PkkF/XVz0+yrZ27LUla/fgk97b6Y0kW9PVZ0V7jhSQrjtSbliRN3VSuKO4ELh5XuwF4uKrOAR5uz0lyLrAcOK/1+XaSOa3P7cAq4Jz2GBtzJfBGVZ0N3Arc0sY6FbgJ+BywGLipP5AkSdNj0qCoqr8Fdo8rXwKsa8frgEv76vdU1btV9SKwE1ic5AzgY1X1aFUVcNe4PmNj3QcsbVcbFwGbqmp3Vb0BbOL9gSVJOsrmHmK/garaBVBVu5J8otXnAT/oazfaaj9vx+PrY31ebmPtTfImcFp/fYI++0myit7VCgMDA4yMjBzi25KOrj179vj51KxzqEFxIJmgVh31Q+2zf7FqLbAWYHBwsIaGhiadqDQTRkZG8POp2eZQ73p6rS0n0b6+3uqjwJl97eYDr7b6/Anq+/VJMhc4md5S14HGkiRNo0MNigeAsbuQVgD399WXtzuZzqK3af14W6Z6K8kFbf/hqnF9xsa6DHik7WNsBJYlOaVtYi9rNUnSNJp06SnJemAIOD3JKL07kb4BbEiyEngJuBygqrYn2QA8B+wFrqmqfW2oq+ndQXUC8FB7AAwDdyfZSe9KYnkba3eSm4EnWruvV9X4TXVJ0lE2aVBU1ZUHOLX0AO3XAGsmqG8FFk1Qf4cWNBOcuwO4Y7I5SpKOHn8zW5LUyaCQJHUyKCRJnQwKSVIng0KS1MmgkCR1MigkSZ0MCklSJ4NCktTJoJAkdTIoJEmdDApJUieDQpLUyaCQJHUyKCRJnQwKSVIng0KS1MmgkKbB+vXrWbRoEUuXLmXRokWsX79+pqckTdmkfwpV0uFZv349q1evZnh4mH379jFnzhxWrlwJwJVXHugvDUvHDq8opKNszZo1DA8Ps2TJEubOncuSJUsYHh5mzZr3/Wl56ZhkUEhH2Y4dO7jwwgv3q1144YXs2LFjhmYkHRyDQjrKFi5cyJYtW/arbdmyhYULF87QjKSD4x6FdJStXr2aK664ghNPPJGXXnqJT33qU7z99tt861vfmumpSVNyWFcUSX6cZFuSp5JsbbVTk2xK8kL7ekpf+xuT7EzyfJKL+urnt3F2JrktSVr9+CT3tvpjSRYcznylmfLOO+/wyiuv8N577/HKK6/wzjvvzPSUpCk7EktPS6rqM1U12J7fADxcVecAD7fnJDkXWA6cB1wMfDvJnNbndmAVcE57XNzqK4E3qups4FbgliMwX2laXX/99Zx00kls3LiRTZs2sXHjRk466SSuv/76mZ6aNCVHY4/iEmBdO14HXNpXv6eq3q2qF4GdwOIkZwAfq6pHq6qAu8b1GRvrPmDp2NWGNFuMjo6ybt26/e56WrduHaOjozM9NWlKDnePooDvJyngP1fVWmCgqnYBVNWuJJ9obecBP+jrO9pqP2/H4+tjfV5uY+1N8iZwGvCz/kkkWUXvioSBgQFGRkYO821JR9bTTz/Ncccdx549exgZGeHpp58G8LOqWeFwg+ILVfVqC4NNSf6uo+1EVwLVUe/qs3+hF1BrAQYHB2toaKhz0tJ0mj9/PjfccAP79u37RW3OnDnMnz8fP6uaDQ5r6amqXm1fXwe+CywGXmvLSbSvr7fmo8CZfd3nA6+2+vwJ6vv1STIXOBnYfThzlqZbkv1CAmDfvn24iqrZ4pCDIsmJST46dgwsA54FHgBWtGYrgPvb8QPA8nYn01n0Nq0fb8tUbyW5oO0/XDWuz9hYlwGPtH0MadZ4+eWXD6ouHWsOZ+lpAPhu+6loLvBfquq/JXkC2JBkJfAScDlAVW1PsgF4DtgLXFNVYz9mXQ3cCZwAPNQeAMPA3Ul20ruSWH4Y85Vm1De/+U3OPfdcnnvuOa677rqZno40Zfmg/YA+ODhYW7dunelpSL+QhCT0f6+NPf+gff9p9kryZN+vOezH/8JDmgbjA8GA0GxiUEiSOhkUkqROBoUkqZNBIUnqZFBIkjoZFJKkTgaFJKmTQSFJ6mRQSJI6GRSSpE4GhSSpk0EhSepkUEiSOhkUkqROBoUkqZNBIUnqZFBIkjoZFJKkTgaFJKmTQSFJ6mRQSJI6GRSSpE6zIiiSXJzk+SQ7k9ww0/ORpA+TYz4okswB/gT4deBc4Mok587srCTpw2PuTE9gChYDO6vqRwBJ7gEuAZ6b0VnpA+dX/+D7vPl/fz7l9j+55YuH/ZpJJm3zS7/7vYMa8+QTjuPpm5Yd6pSk95kNQTEPeLnv+Sjwuf4GSVYBqwAGBgYYGRmZtsnpg+O9Bdfx0YNov+jORUdtLvs7uNXW94CRkT8+OlPRh9JsCIqJfuSq/Z5UrQXWAgwODtbQ0NA0TEsfNNvYdlTG7bpqqKoDnpOOFcf8HgW9K4gz+57PB16doblIB+1AYWBIaLaYDUHxBHBOkrOS/CNgOfDADM9JOihVRVWxefPmXxxLs8Uxv/RUVXuTfAXYCMwB7qiq7TM8LUn60DjmgwKgqh4EHpzpeUjSh9FsWHqSJM0gg0KS1MmgkCR1MigkSZ3yQbtNL8lPgZ/M9DykAzgd+NlMT0KawC9V1ccnOvGBCwrpWJZka1UNzvQ8pIPh0pMkqZNBIUnqZFBI02vtTE9AOljuUUiSOnlFIUnqZFBIkjoZFJKkTgaF1CT5n+3rgiT/ZqbnA5Dkk0num+l56MPNzWxpnCRDwL+vqi/O8DzmVtXemZyDBF5RSL+QZE87/AbwL5I8leTfJZmT5D8keSLJM0m+3NoPJfmbJBuS/K8k30jyb5M8nmRbkl/ueK07k/xpkv/e+n6x1X87yV8k+a/A99vVzbPt3Jwkf9jGfibJta1+fpvHk0k2JjnjqP5D6UNnVvzhImma3UDfFUWSVcCbVfXPkxwP/I8k329tfxVYCOwGfgT8WVUtTvJV4Frgax2vswD4l8AvA5uTnN3qnwd+pap2J1nQ134VcBbw2faXH09Nchzwx8AlVfXTJFcAa4AvHd4/gfQPDAppcsuAX0lyWXt+MnAO8P+AJ6pqF0CS/w2MBcg2YMkk426oqveAF5L8CPhnrb6pqnZP0P7XgD8dW45qQbIIWARsSgK9Pxe86xDeo3RABoU0uQDXVtXG/Yq9vYx3+0rv9T1/j8m/v8ZvEI49f7tjHuP7BNheVZ+f5LWkQ+YehfR+bwEf7Xu+Ebi6LfOQ5NNJTjwCr3N5ko+0vYx/Cjw/SfvvA7+TZG6bx6mtz8eTfL7Vjkty3hGYm/QLXlFI7/cMsDfJ08CdwLfo7Sf8ML31nZ8Clx6B13ke+BtgAPidqnqnLR8dyJ8BnwaeSfJz4DtV9Z/akthtSU6m9z39H4HtR2B+EuDtsdKMSHIn8L2q8nckdMxz6UmS1MmlJ+koSrIauHxc+S+q6rdnYDrSIXHpSZLUyaUnSVIng0KS1MmgkCR1MigkSZ3+P1n9vDFcyJYPAAAAAElFTkSuQmCC\n"},"metadata":{"needs_background":"light"}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"train = train[(train['item_cnt_day'] >= 0) &\n              (train['item_cnt_day'] <= 800) &\n              (train['item_price'] >= 0) &\n              (train['item_price'] <= 100000)]\n\nprint(f'Quantidade de elementos retirados: {old_size - train.shape[0]}')","execution_count":8,"outputs":[{"output_type":"stream","text":"Quantidade de elementos retirados: 7360\n","name":"stdout"}]},{"metadata":{},"cell_type":"markdown","source":"Uma variável interessante de ser utilizada na análise é a **categoria** dos itens, pois agrupa os produtos em relação a suas características, que podem ter alguma relação as vendas. Esse atributo foi adicionado utilizando a função [**merge**](https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.merge.html) do pandas."},{"metadata":{"trusted":true},"cell_type":"code","source":"# Adiciona a categoria dos itens ao dataframe\ntrain = pd.merge(train, items, on='item_id', how='left')\ntest = pd.merge(test, items, on='item_id', how='left')","execution_count":9,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"Manipular dataframes com grandes quantidades de informação (quase 3 milhões de elementos no dataframe de treino) pode ser um problema, pois torna o pré-processamento e o processamento computacionalmente custosos, necessitando de uma grande quantidade de memória para manipular os dados. Para amenizar esse problema, utilizaremos de uma função de redução de memória no dataframe principal, que procura armazenar os dados numéricos em variáveis com menos precisão, mas que ocupam menos espaço em memória. A função pode ser vista abaixo e encontrada neste [link](https://www.kaggle.com/gemartin/load-data-reduce-memory-usage), de onde foi retirada."},{"metadata":{"trusted":true},"cell_type":"code","source":"def reduce_mem_usage(df):\n    start_mem = df.memory_usage().sum() / 1024**2\n    print('Memória utilizada pelo dataframe: {:.2f} MB.'.format(start_mem))\n    for col in df.columns:\n        col_type = df[col].dtype\n        if col_type != object:\n            c_min = df[col].min()\n            c_max = df[col].max()\n            if str(col_type)[:3] == 'int':\n                if c_min > np.iinfo(np.int8).min and c_max < np.iinfo(np.int8).max:\n                    df[col] = df[col].astype(np.int8)\n                elif c_min > np.iinfo(np.int16).min and c_max < np.iinfo(np.int16).max:\n                    df[col] = df[col].astype(np.int16)\n                elif c_min > np.iinfo(np.int32).min and c_max < np.iinfo(np.int32).max:\n                    df[col] = df[col].astype(np.int32)\n                elif c_min > np.iinfo(np.int64).min and c_max < np.iinfo(np.int64).max:\n                    df[col] = df[col].astype(np.int64)\n            else:\n                if c_min > np.finfo(np.float16).min and c_max < np.finfo(np.float16).max:\n                    df[col] = df[col].astype(np.float16)\n                elif c_min > np.finfo(np.float32).min and c_max < np.finfo(np.float32).max:\n                    df[col] = df[col].astype(np.float32)\n                else:\n                    df[col] = df[col].astype(np.float64)\n        else:\n            df[col] = df[col].astype('category')\n    end_mem = df.memory_usage().sum() / 1024**2\n    print('Memória utilizada após a otimização: {:.2f} MB.'.format(end_mem))\n    print('Uso de memória reduzido em {:.1f}%.'.format(100 * (start_mem - end_mem) / start_mem))\n\n    return df","execution_count":10,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"Após as modificações feitas, utilizamos a função de redução de memória no dataframe para otimizar os próximos processos."},{"metadata":{"trusted":true},"cell_type":"code","source":"# Reduz a memória utilizada\ntrain = reduce_mem_usage(train)","execution_count":11,"outputs":[{"output_type":"stream","text":"Memória utilizada pelo dataframe: 156.40 MB.\nMemória utilizada após a otimização: 47.48 MB.\nUso de memória reduzido em 69.6%.\n","name":"stdout"}]},{"metadata":{},"cell_type":"markdown","source":"O dataframe utilizado descreve ocorrências de vendas, mas não a ausência delas, não se tem a informação de quando ocorreram zero vendas para certos itens em certas lojas, e sem essa informação o regressor ficará enviesado por apenas possuir informações de quando as vendas ocorrem, assumindo que pelo menos um item será vendido em todas as predições, mesmo sendo possível que nenhum item seja vendido naquele dia.\n\nPara resolver esse problema, precisamos gerar esses dados de quando não existem vendas, que são muitos, pois precisamos inserir zeros para cada combinação de item e loja que não possui informação de venda. Abaixo temos a criação desses novos dados, primeiro iniciamos a variável **grid** e nela geramos todas essas combinações de itens e lojas para um certo mês, para isso utilizamos o método [**unique**](https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.Series.unique.html) para identificar os valores únicos de item e loja, depois da utilizamos da função [**product**](https://docs.python.org/2/library/itertools.html) para gerar as combinações. Por fim, utilizamos a função [**vstack**](https://numpy.org/doc/stable/reference/generated/numpy.vstack.html) para gerar um dataframe novo, onde as combinações por mês estão empilhadas."},{"metadata":{"trusted":true},"cell_type":"code","source":"from itertools import product\n\nindex_cols = ['shop_id', 'item_id', 'date_block_num']\n# Cria uma combinação de cada loja/item daquele mês.\ngrid = []\nfor block_num in train['date_block_num'].unique():\n    cur_shops = train.loc[train['date_block_num'] == block_num, 'shop_id'].unique()\n    cur_items = train.loc[train['date_block_num'] == block_num, 'item_id'].unique()\n    grid.append(np.array(list(product(*[cur_shops, cur_items, [block_num]])), dtype='int32'))\ngrid = pd.DataFrame(np.vstack(grid), columns=index_cols, dtype=np.int32)\n\nprint(f'\\nQuantidade de elementos: {grid.shape[0]}\\n')\n\nprint(grid)","execution_count":12,"outputs":[{"output_type":"stream","text":"\nQuantidade de elementos: 10884508\n\n          shop_id  item_id  date_block_num\n0              59    22154               0\n1              59     2552               0\n2              59     2554               0\n3              59     2555               0\n4              59     2564               0\n...           ...      ...             ...\n10884503       21     7635              33\n10884504       21     7638              33\n10884505       21     7640              33\n10884506       21     7632              33\n10884507       21     7440              33\n\n[10884508 rows x 3 columns]\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"# Reduz a memória utilizada\ngrid = reduce_mem_usage(grid)","execution_count":13,"outputs":[{"output_type":"stream","text":"Memória utilizada pelo dataframe: 124.56 MB.\nMemória utilizada após a otimização: 41.52 MB.\nUso de memória reduzido em 66.7%.\n","name":"stdout"}]},{"metadata":{},"cell_type":"markdown","source":"Uma vez que o algoritmo possui o intuito de prever vendas de um mês inteiro, e os dados que são fornecidos são de vendas diárias, optamos por agrupar esses dados por mês e não utilizar a informação do dia exato da venda. Essa abordagem é menos custosa computacionalmente, pois diminui a quantidade de dados ao agrupá-los, e é uma abordagem que faz sentido pensando no objetivo final da competição, que é predizer vendas em todo um mês. Para isso utilizamos a função [**groupby**](https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.groupby.html) para agrupar as vendas de meses iguais, somando as vendas, tirando a média dos preços e armazenando essa nova informação no dataframe **sales_m**."},{"metadata":{"trusted":true},"cell_type":"code","source":"sales_m = train.groupby(index_cols).agg({'item_cnt_day': 'sum',\n                                         'item_price': np.mean}).reset_index()\n\nprint(sales_m)","execution_count":14,"outputs":[{"output_type":"stream","text":"         shop_id  item_id  date_block_num  item_cnt_day  item_price\n0              0       30               1          31.0       265.0\n1              0       31               1          11.0       434.0\n2              0       32               0           6.0       221.0\n3              0       32               1          10.0       221.0\n4              0       33               0           3.0       347.0\n...          ...      ...             ...           ...         ...\n1608219       59    22164              27           2.0       699.0\n1608220       59    22164              30           1.0       699.0\n1608221       59    22167               9           1.0       299.0\n1608222       59    22167              11           2.0       299.0\n1608223       59    22167              17           1.0       299.0\n\n[1608224 rows x 5 columns]\n","name":"stdout"}]},{"metadata":{},"cell_type":"markdown","source":"Depois, utilizamos a função **merge** para unir as novas combinações de item/loja da variável **grid** no novo dataframe **sales_m**, e preenchemos eles com zeros."},{"metadata":{"trusted":true},"cell_type":"code","source":"sales_m = pd.merge(grid, sales_m, on=index_cols, how='left').fillna(0)\n\nprint(f'Quantidade de elementos: {sales_m.shape[0]}\\n')\nprint(sales_m)","execution_count":15,"outputs":[{"output_type":"stream","text":"Quantidade de elementos: 10884508\n\n          shop_id  item_id  date_block_num  item_cnt_day  item_price\n0              59    22154               0           1.0       999.0\n1              59     2552               0           0.0         0.0\n2              59     2554               0           0.0         0.0\n3              59     2555               0           0.0         0.0\n4              59     2564               0           0.0         0.0\n...           ...      ...             ...           ...         ...\n10884503       21     7635              33           0.0         0.0\n10884504       21     7638              33           0.0         0.0\n10884505       21     7640              33           0.0         0.0\n10884506       21     7632              33           0.0         0.0\n10884507       21     7440              33           0.0         0.0\n\n[10884508 rows x 5 columns]\n","name":"stdout"}]},{"metadata":{},"cell_type":"markdown","source":"Adicionamos também no novo dataframe as categorias dos itens."},{"metadata":{"trusted":true},"cell_type":"code","source":"sales_m = pd.merge(sales_m, items, on='item_id', how='left')","execution_count":16,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Novos atributos\n\nAgora, criamos novos atributos de média e soma da quantidade de itens e de média do preço, isso para cada item, loja e categoria diferente. Essa abordagem permite que o algoritmo receba informações mais gerais a respeito dos preços e do número de vendas dos itens, das lojas e das categorias, fazendo que o programa consiga analisar a situação das vendas recentes analisando se um produto ou loja ficou ou saiu de tendência."},{"metadata":{"trusted":true},"cell_type":"code","source":"for type_id in ['item_id', 'shop_id', 'item_category_id']:\n    for column_id, aggregator, aggtype in [('item_price', np.mean, 'avg'),\n                                           ('item_cnt_day', np.sum, 'sum'),\n                                           ('item_cnt_day', np.mean, 'avg')]:\n\n        # Gera os novos atributos e renomeia as colunas\n        mean_df = train.groupby([type_id, 'date_block_num']).aggregate(aggregator).reset_index()[[column_id, type_id, 'date_block_num']]\n        mean_df.columns = [type_id+'_'+aggtype+'_'+column_id, type_id, 'date_block_num']\n        \n        # Une os novos atributos ao dataframe original\n        sales_m = pd.merge(sales_m, mean_df, on=['date_block_num', type_id], how='left')\n    \nprint(sales_m.columns)","execution_count":17,"outputs":[{"output_type":"stream","text":"Index(['shop_id', 'item_id', 'date_block_num', 'item_cnt_day', 'item_price',\n       'item_category_id', 'item_id_avg_item_price',\n       'item_id_sum_item_cnt_day', 'item_id_avg_item_cnt_day',\n       'shop_id_avg_item_price', 'shop_id_sum_item_cnt_day',\n       'shop_id_avg_item_cnt_day', 'item_category_id_avg_item_price',\n       'item_category_id_sum_item_cnt_day',\n       'item_category_id_avg_item_cnt_day'],\n      dtype='object')\n","name":"stdout"}]},{"metadata":{},"cell_type":"markdown","source":"### Lag features\n\nAs **lag features** são abordagens muito utilizadas em problemas de predição temporal, pois os transforma em problemas supervisionados. A ideia é tentar prever valores em um tempo **t** dadas as informações dos tempos **t-1**, **t-2**, ... Essas informações auxiliam o algoritmo a prever vendas em um certo mês utilizando informações de vendas nos meses passados.\nPara gerar as lag features, criamos novos atributos temporais com informações de meses passados para cada um dos elementos, a partir de uma janela temporal que define quais meses anteriores serão analisados. A janela analisa os 6 últimos meses antes da predição, e este mesmo mês no ano passado.\nCaso queira mais informações sobre lag features [clique aqui]( https://machinelearningmastery.com/basic-feature-engineering-time-series-data-python/).\nAbaixo temos a implementação de uma função que gera as lag features no dataframe."},{"metadata":{"trusted":true},"cell_type":"code","source":"# Definimos quais variáveis serão analisadas dos meses passados\nlag_variables = list(sales_m.columns[6:])+['item_cnt_day']\n# Definimos a janela temporal\nlags = [1, 2, 3, 4, 5, 6, 12]\n    \ndef lag_features(df, sales_m):\n    for lag in lags:\n        # Cria-se uma cópia do sales_m\n        sales_new_df = sales_m.copy()\n        # Translada os meses adicionando o valor lag a eles\n        sales_new_df.date_block_num += lag\n        # Adiciona as novas colunas com as variáveis temporais\n        sales_new_df = sales_new_df[index_cols+lag_variables]\n        sales_new_df.columns = index_cols + [f'{lag_feat}_lag_{lag}'\n                                             for lag_feat in lag_variables]\n        # Une o dataset original com o com os atributos temporais, usando de referência\n        # o identificador do item e da loja, e o mês de venda (index_cols).\n        df = pd.merge(df, sales_new_df, on=index_cols, how='left')\n\n    return df\n\n\nlag_train = sales_m.copy()\nlag_train = lag_features(lag_train, sales_m)\n\nprint(f'Quantidade de atributos no dataframe: {lag_train.shape[1]}\\n')\nprint(lag_train.columns)","execution_count":18,"outputs":[{"output_type":"stream","text":"Quantidade de atributos no dataframe: 85\n\nIndex(['shop_id', 'item_id', 'date_block_num', 'item_cnt_day', 'item_price',\n       'item_category_id', 'item_id_avg_item_price',\n       'item_id_sum_item_cnt_day', 'item_id_avg_item_cnt_day',\n       'shop_id_avg_item_price', 'shop_id_sum_item_cnt_day',\n       'shop_id_avg_item_cnt_day', 'item_category_id_avg_item_price',\n       'item_category_id_sum_item_cnt_day',\n       'item_category_id_avg_item_cnt_day', 'item_id_avg_item_price_lag_1',\n       'item_id_sum_item_cnt_day_lag_1', 'item_id_avg_item_cnt_day_lag_1',\n       'shop_id_avg_item_price_lag_1', 'shop_id_sum_item_cnt_day_lag_1',\n       'shop_id_avg_item_cnt_day_lag_1',\n       'item_category_id_avg_item_price_lag_1',\n       'item_category_id_sum_item_cnt_day_lag_1',\n       'item_category_id_avg_item_cnt_day_lag_1', 'item_cnt_day_lag_1',\n       'item_id_avg_item_price_lag_2', 'item_id_sum_item_cnt_day_lag_2',\n       'item_id_avg_item_cnt_day_lag_2', 'shop_id_avg_item_price_lag_2',\n       'shop_id_sum_item_cnt_day_lag_2', 'shop_id_avg_item_cnt_day_lag_2',\n       'item_category_id_avg_item_price_lag_2',\n       'item_category_id_sum_item_cnt_day_lag_2',\n       'item_category_id_avg_item_cnt_day_lag_2', 'item_cnt_day_lag_2',\n       'item_id_avg_item_price_lag_3', 'item_id_sum_item_cnt_day_lag_3',\n       'item_id_avg_item_cnt_day_lag_3', 'shop_id_avg_item_price_lag_3',\n       'shop_id_sum_item_cnt_day_lag_3', 'shop_id_avg_item_cnt_day_lag_3',\n       'item_category_id_avg_item_price_lag_3',\n       'item_category_id_sum_item_cnt_day_lag_3',\n       'item_category_id_avg_item_cnt_day_lag_3', 'item_cnt_day_lag_3',\n       'item_id_avg_item_price_lag_4', 'item_id_sum_item_cnt_day_lag_4',\n       'item_id_avg_item_cnt_day_lag_4', 'shop_id_avg_item_price_lag_4',\n       'shop_id_sum_item_cnt_day_lag_4', 'shop_id_avg_item_cnt_day_lag_4',\n       'item_category_id_avg_item_price_lag_4',\n       'item_category_id_sum_item_cnt_day_lag_4',\n       'item_category_id_avg_item_cnt_day_lag_4', 'item_cnt_day_lag_4',\n       'item_id_avg_item_price_lag_5', 'item_id_sum_item_cnt_day_lag_5',\n       'item_id_avg_item_cnt_day_lag_5', 'shop_id_avg_item_price_lag_5',\n       'shop_id_sum_item_cnt_day_lag_5', 'shop_id_avg_item_cnt_day_lag_5',\n       'item_category_id_avg_item_price_lag_5',\n       'item_category_id_sum_item_cnt_day_lag_5',\n       'item_category_id_avg_item_cnt_day_lag_5', 'item_cnt_day_lag_5',\n       'item_id_avg_item_price_lag_6', 'item_id_sum_item_cnt_day_lag_6',\n       'item_id_avg_item_cnt_day_lag_6', 'shop_id_avg_item_price_lag_6',\n       'shop_id_sum_item_cnt_day_lag_6', 'shop_id_avg_item_cnt_day_lag_6',\n       'item_category_id_avg_item_price_lag_6',\n       'item_category_id_sum_item_cnt_day_lag_6',\n       'item_category_id_avg_item_cnt_day_lag_6', 'item_cnt_day_lag_6',\n       'item_id_avg_item_price_lag_12', 'item_id_sum_item_cnt_day_lag_12',\n       'item_id_avg_item_cnt_day_lag_12', 'shop_id_avg_item_price_lag_12',\n       'shop_id_sum_item_cnt_day_lag_12', 'shop_id_avg_item_cnt_day_lag_12',\n       'item_category_id_avg_item_price_lag_12',\n       'item_category_id_sum_item_cnt_day_lag_12',\n       'item_category_id_avg_item_cnt_day_lag_12', 'item_cnt_day_lag_12'],\n      dtype='object')\n","name":"stdout"}]},{"metadata":{},"cell_type":"markdown","source":"Com os novos atributos temporais prontos, precisamos preencher os valores ausentes, abaixo a função **fillNaN** trocamos todos os valores NaN gerados acima por zero quando se trata da quantidade de itens vendidas já que quando não há informação indica que nada foi vendido daquele item na loja, ou pela mediana quando se trata dos preços já que durante um único mês pode existir a variação dos preços dos itens, e foi decidido colocar a mediana para representar o valor durante o mês inteiro."},{"metadata":{"trusted":true},"cell_type":"code","source":"def fillNaN(df):\n    for feat in df.columns:\n        if 'item_cnt' in feat:\n            df[feat] = df[feat].fillna(0)\n        elif 'item_price' in feat:\n            df[feat] = df[feat].fillna(df[feat].median())\n\nfillNaN(lag_train)","execution_count":19,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"Por utilizar a abordagem das lag features, os primeiros doze meses do dataframe não possuem as informações necessárias para gerar todas as variáveis temporais, por isso eles são retirados do dataframe. Depois, para lidar melhor com a [sazonalidade]( https://www.devmedia.com.br/data-mining-na-pratica-time-series/5414) anual de venda dos produtos, utilizamos a abordagem de gerar dois novos atributos, um de ano, com um valor entre 0 e 2, e outro de mês, com um valor entre 0 e 11, dessa forma é possível isolar características que dizem respeito a certos períodos do ano."},{"metadata":{"trusted":true},"cell_type":"code","source":"def date_attr(df):\n    df['month'] = df['date_block_num'] % 12\n    df['year'] = df['date_block_num'] // 12\n    df.drop(columns='date_block_num', inplace=True)\n    return df\n\n# Retira os meses do primeiro ano\nlag_train = lag_train[lag_train['date_block_num'] > 12]\n# Gera novos atributos de ano e mês\nlag_train = date_attr(lag_train)","execution_count":20,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# Retira os atributos que não serão utilizados na análise,\n# como os atributos utilizados para gerar as lag features.\ncols_to_drop = lag_variables[:-1] + ['item_price']\nlag_train.drop(columns=cols_to_drop, inplace=True)\n\n# Reduz a memória utilizada\nlag_train = reduce_mem_usage(lag_train)","execution_count":21,"outputs":[{"output_type":"stream","text":"Memória utilizada pelo dataframe: 941.79 MB.\nMemória utilizada após a otimização: 901.35 MB.\nUso de memória reduzido em 4.3%.\n","name":"stdout"}]},{"metadata":{},"cell_type":"markdown","source":"Com o dataset de treino completo, precisamos inserir essas variáveis temporais nos testes, já que apenas possuímos a informação de qual item em qual loja desejamos prever. Para isso, inicializamos a data como o 35º mês (valor 34), e utilizamos das mesmas funções implementadas ao gerar os casos de treino para gerar as lag_features, preencher os dados vazios gerados e adicionar os novos dois atributos de mês e ano."},{"metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"cell_type":"code","source":"test['date_block_num'] = 34\n\ntest = lag_features(test, sales_m)\nfillNaN(test)\ntest = date_attr(test)","execution_count":22,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"O dataframe **lag_train** possui as informações para o treino do regressor, todos os atributos preditivos mais a coluna **item_cnt_day**, que representa o número de vendas, informação que se deseja prever com esses testes.\n\nO dataframe **test** possui os mesmos atributos preditivos que o dataframe de treino, com informações do 35º mês, porém sem a coluna **item_cnt_day** que é o que se deseja prever.\n\nUm teste abaixo foi feito para confirmar que os dataframes possuem a mesma estrutura."},{"metadata":{"trusted":true},"cell_type":"code","source":"# Verificando se o treino e o teste possuem as mesmas colunas\n_test = set(test.columns)\n_train = set(lag_train.drop(columns='item_cnt_day').columns)\nassert _test==_train","execution_count":23,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Processamento\n\nApós o pré-processamento feito no dataframe, ele está pronto para ser utilizado pelos algoritmos regressores. Aqui testaremos com dois deles, o [Random Forest]( https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestRegressor.html) e o [XGBoost]( https://xgboost.readthedocs.io/en/latest/python/python_api.html?highlight=matrix#xgboost.DMatrix), para que seja possível analisarmos a diferença dos resultados obtidos utilizando dois algoritmos diferentes.\n\n### Random Forest\n\nApós todas as análises feitas, e todo tratamento com a base, estamos enfim prontos para processar e estabelecer nossas predições. Abaixo temos a finalização do projeto, aqui inicialmente usamos uma regressão com floresta aleatória criando 10 árvores, com n_jobs = -1, para usar todas as unidades centrais de processamento, já que a base, mesmo com funções para diminuir o custo de memória ainda é muito pesado foi necessário reduzir o número de árvores para dez o que pelos testes feitos não afetaram tanto na precisão da predição. Treinamos a floresta com a base trabalhada, e transformamos o exemplo de submissão da base na nossa predição. Além disso devemos para a submissão devemos limitar as vendas da predição de 0 a 20 para enfim transformar em um arquivo csv e submeter."},{"metadata":{"trusted":true},"cell_type":"code","source":"from sklearn.ensemble import RandomForestRegressor\n\nregressor = RandomForestRegressor(n_estimators=10, n_jobs = -1)\n\nregressor.fit(lag_train.drop(columns=['item_cnt_day']),\n              lag_train['item_cnt_day'])\n\npred = regressor.predict(test)\n\npred = pred.clip(0, 20)\n\n\nsubmission['item_cnt_month'] = pred\n\nsubmission.to_csv('submission_rf.csv', index=False)","execution_count":24,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# Feature importance\nfeat_importances = pd.Series(regressor.feature_importances_, index=lag_train.drop(columns=['item_cnt_day']).columns)\nfeat_importances.nlargest(10).plot(kind='barh')","execution_count":25,"outputs":[{"output_type":"execute_result","execution_count":25,"data":{"text/plain":"<matplotlib.axes._subplots.AxesSubplot at 0x7fb950919550>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<Figure size 432x288 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"metadata":{},"cell_type":"markdown","source":"Como pode-se ver acima foi mostrado o gráfico com a feature importances que indica quais atributos mais influenciaram na predição, nesse caso o programa definiu que a informações mais importantes são a venda do mês passado, o item que está sendo vendido e a loja onde ele está sendo vendido, o que são resultados bem interessantes ao pensar que a venda do mês passado seja a informação mais importante por ser a mais recente."},{"metadata":{},"cell_type":"markdown","source":"A abordagem utilizando Random Forest alcançou um erro de 1.10"},{"metadata":{},"cell_type":"markdown","source":"### XGBoost\n\nDepois de tratar os dados da predição, agora sim podemos processá-la. Primeiro é claro treinamos com os meses passados o xgboost com os parâmetros abaixo. Depois fazemos a predição de vendas do xgboost, e criamos o arquivo padrão para envio."},{"metadata":{},"cell_type":"markdown","source":"Os parâmetros foram escolhidos em sua maioria a partir da análise do dataset feito em outros notebooks da competição principalmente [esse]( https://www.kaggle.com/alessandrosolbiati/using-xgboost-for-time-series-prediction-top-20\n), alguns foram testes que nós mesmos fizemos, que envolve em sua maioria reduzir o uso de memória. O parâmetro **max_depth** foi definido em 10 para dar uma maior liberdade de crescimento do programa que o valor default de 6, foi decidido colocar a **verbosity** em 0 para deixar o output mais limpo, e **eval_metric** foi decidido como forma de auto avaliação do algoritmo para aprendizado ser igual à da competição, para os outros parâmetros foi decidido manter o valor default do programa depois dos testes darem bons resultados."},{"metadata":{"trusted":true},"cell_type":"code","source":"param = {'max_depth': 10,\n         'subsample': 1,\n         'min_child_weight': 1,\n         'eta': 0.3,\n         'num_round': 1000,\n         'eval_metric': 'rmse',\n         'verbosity': 0}","execution_count":26,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"import xgboost as xgb\n\nxgbtrain = xgb.DMatrix(lag_train.drop(columns=['item_cnt_day']),\n                       lag_train['item_cnt_day'])\n\nbst = xgb.train(param, xgbtrain)\n\nxgbpredict = xgb.DMatrix(test)\npred = bst.predict(xgbpredict)\npred = pred.clip(0, 20)\n\nsubmission['item_cnt_month'] = pred\n\nsubmission.to_csv('submission_xgboost.csv', index=False)","execution_count":27,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# Feature importance\nx = xgb.plot_importance(bst)\nx.figure.set_size_inches(10, 30)","execution_count":28,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 720x2160 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"metadata":{},"cell_type":"markdown","source":"Aqui o algoritmo percebe uma importância muito grande para qual loja estamos falando, seguida das vendas do mês passado, o item sendo analisado, o mês do ano em que estamos, e a média de preços dos itens vendidos no mês passado. A enorme importância das lojas é intuitiva, já que existem lojas que nunca vendem nada. O item também faz sentido, já que produtos são naturalmente mais vendidos e necessários que outros.\n\nMas o interessante a se notar aqui é talvez, a diferença entre as importâncias de atributos dado por ambos os modelos. Que se difere em muitos pontos, mas outros são idênticos, talvez seja esse o motivo pelo qual o resultado foi diferente, e mostram também a maior precisão do xgboost para o dataset final."},{"metadata":{},"cell_type":"markdown","source":"A abordagem utilizando XGBoost alcançou um erro de 0.93, e garantindo a posição 2603º no rank da competição no dia 30/06/2020."},{"metadata":{},"cell_type":"markdown","source":"## Conclusão\n\nA abordagem utilizando lag features foi eficaz e trouxe resultados consideráveis, alcançamos um erro rmse de 0.93 utilizando xgboost, que demora por volta de 30 minutos para treinar e predizer as vendas, e um erro rmse de 1.10 utilizando a random forest, que não conseguiu resultados tão bons quanto o xgboost e levava por volta de 1 hora para treinar e predizer as vendas. Ambas as abordagens alcançaram resultados melhores que os valores iniciais do arquivo de submissão, que chuta todas as predições como 0.5, que alcançou 1.23 de erro."},{"metadata":{},"cell_type":"markdown","source":"## Evolução do programa\nDurante a criação desse notebook foram criadas várias abordagens do problema, que ao longo do tempo foram naturalmente evoluindo. Abaixo descrevemos como essa evolução se deu. \n\n### versão 0\n\nPrimeiramente nós começamos apenas colocando o data set no programa para ver os resultados, para assim entendermos melhor as notas da competição. Nos deparamos então com o primeiro problema do dataset, ele vinha com os dados de dias e o target era um mês inteiro, felizmente esse foi um problema que foi facilmente resolvido por juntar as informações dos dias do mês apenas no mês.\n\n### versão 1\n\nO que nos levou ao segundo problema pelo formato do dataset, nós só temos dados de quando houveram vendas. Isso foi um grande obstáculo pois isso indica para o algoritmo, que nós sempre temos vendas o que é muito ruim para análise, já que a maioria das vezes não existem vendas. Então nós pegamos uma solução simples para esse problema, inserir os zeros manualmente, porém isso  veio com um custo, um alto consumo de memória que se não parasse o algoritmo em fase de manipular os dados, o programa morria no treino.\n\n### versão 2\n\nApós isso criamos outra abordagem a de utilizar as categorias, que dessa forma o programa poderia adicionar os zeros e ainda treinar de uma forma tranquila. Porém apesar de ter um resultado melhor que o primeiro por uma grande margem e funcionar diferentemente da segunda abordagem, nós não estamos satisfeitos com simplesmente analisar os itens por categoria e dar resultados iguais a todos os itens, de uma categoria, já que isso é de certa forma contraproducente, já que é mais uma super generalização do comportamento do que uma solução de fato. Então nós fomos a terceira abordagem e o início da implementação de xgboost.\n\n### versão 3\n\nDiferentemente das outras abordagens nós tentamos ver o problema em uma visão temporal onde cada mês anterior se torna um atributo para o próximo e esse se torna o target. Essa foi por muito tempo nosso melhor resultado em xgboost porém ele não se mostrou tão efetivo quanto o método anterior para Random Forest levando o grupo a tentar diferente.\n\n### versão 4\n\nNesse momento nós tentamos dar uma olhada de volta na abordagem 2 para tentarmos melhora-la e encontramos a [reduce_mem_usage]( https://www.kaggle.com/gemartin/load-data-reduce-memory-usage) uma função muito útil que utiliza da redução de variável para reduzir o consumo de memoria. Tal função resultou em uma redução de quase 80% do uso de memória deixando a abordagem 1 viável novamente e com isso tivemos um resultado melhor.\n\n### versão 5\n\nAqui percebemos que ambas as abordagens que testamos se tornaram possíveis, e obtiveram bons resultados. Separadamente. Então tentamos ver se ambos juntas poderiam melhorar o nosso resultado. A ideia era utilizar as categorias e a redução de memória para colocar todos os 0. Mas a suposição se mostrou errada, o resultado se tornou pior que ambos separados. Provavelmente o algoritmo se enviesa nessa união, além de ter se tornado mais custoso.\n\n### versão 6\n\nPor fim resolvemos tentar implementar novamente o viés temporal no algoritmo, essa versão é versão do notebook acima, com os resultados muito melhores do que os anteriores, tanto para Random Forest, quanto para Xgboost. Nesta versão usamos também partes das versões anteriores para melhorar a predição com aspectos chave de cada uma."}],"metadata":{"kernelspec":{"name":"python3","display_name":"Python 3","language":"python"},"language_info":{"name":"python","version":"3.7.6","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"}},"nbformat":4,"nbformat_minor":4}