3.1 - Train Neural Net, one layer (32).ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "303fd31b",
   "metadata": {},
   "outputs": [],
   "source": [
    "import random\n",
    "\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import tensorflow as tf\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense\n",
    "from tensorflow.keras.utils import to_categorical\n",
    "from tensorflow.keras.callbacks import EarlyStopping\n",
    "from keras.optimizers import adam_v2\n",
    "from sklearn.metrics import log_loss\n",
    "\n",
    "from utils import scoring_function\n",
    "\n",
    "pd.set_option('display.max_rows', 200)\n",
    "pd.set_option('display.max_columns', 100)\n",
    "plt.rcParams['figure.figsize'] = (15, 8)\n",
    "plt.rc('axes', axisbelow=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "e6b9e2f4",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(432587, 40) (144196, 40) (432587,) (144196,)\n"
     ]
    }
   ],
   "source": [
    "X_train = pd.read_feather('X_train_standard.f').set_index('sku')\n",
    "X_val = pd.read_feather('X_val_standard.f').set_index('sku')\n",
    "y_train = pd.read_feather('y_train.f').set_index('sku').iloc[:, 0]\n",
    "y_val = pd.read_feather('y_val.f').set_index('sku').iloc[:, 0]\n",
    "print(X_train.shape, X_val.shape, y_train.shape, y_val.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "48a3edce",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2021-09-20 20:36:09.387879: I tensorflow/core/platform/cpu_feature_guard.cc:142] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations:  AVX2 FMA\n",
      "To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.\n",
      "2021-09-20 20:36:09.803529: I tensorflow/compiler/mlir/mlir_graph_optimization_pass.cc:176] None of the MLIR Optimization Passes are enabled (registered 2)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/300\n",
      "13519/13519 [==============================] - 25s 927us/step - loss: 3.5850 - val_loss: 3.2768\n",
      "Epoch 2/300\n",
      "13519/13519 [==============================] - 15s 1ms/step - loss: 3.2561 - val_loss: 3.1923\n",
      "Epoch 3/300\n",
      "13519/13519 [==============================] - 13s 974us/step - loss: 3.1878 - val_loss: 3.1739\n",
      "Epoch 4/300\n",
      "13519/13519 [==============================] - 12s 865us/step - loss: 3.1712 - val_loss: 3.1607\n",
      "Epoch 5/300\n",
      "13519/13519 [==============================] - 13s 929us/step - loss: 3.1563 - val_loss: 3.1561\n",
      "Epoch 6/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1517 - val_loss: 3.1468\n",
      "Epoch 7/300\n",
      "13519/13519 [==============================] - 12s 891us/step - loss: 3.1458 - val_loss: 3.1445\n",
      "Epoch 8/300\n",
      "13519/13519 [==============================] - 11s 841us/step - loss: 3.1432 - val_loss: 3.1401\n",
      "Epoch 9/300\n",
      "13519/13519 [==============================] - 12s 865us/step - loss: 3.1378 - val_loss: 3.1370\n",
      "Epoch 10/300\n",
      "13519/13519 [==============================] - 11s 805us/step - loss: 3.1389 - val_loss: 3.1353\n",
      "Epoch 11/300\n",
      "13519/13519 [==============================] - 11s 795us/step - loss: 3.1356 - val_loss: 3.1326\n",
      "Epoch 12/300\n",
      "13519/13519 [==============================] - 11s 794us/step - loss: 3.1353 - val_loss: 3.1332\n",
      "Epoch 13/300\n",
      "13519/13519 [==============================] - 12s 903us/step - loss: 3.1324 - val_loss: 3.1308\n",
      "Epoch 14/300\n",
      "13519/13519 [==============================] - 17s 1ms/step - loss: 3.1317 - val_loss: 3.1298\n",
      "Epoch 15/300\n",
      "13519/13519 [==============================] - 17s 1ms/step - loss: 3.1292 - val_loss: 3.1293\n",
      "Epoch 16/300\n",
      "13519/13519 [==============================] - 17s 1ms/step - loss: 3.1293 - val_loss: 3.1275\n",
      "Epoch 17/300\n",
      "13519/13519 [==============================] - 22s 2ms/step - loss: 3.1299 - val_loss: 3.1275\n",
      "Epoch 18/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1266 - val_loss: 3.1276\n",
      "Epoch 19/300\n",
      "13519/13519 [==============================] - 13s 973us/step - loss: 3.1265 - val_loss: 3.1282\n",
      "Epoch 20/300\n",
      "13519/13519 [==============================] - 15s 1ms/step - loss: 3.1244 - val_loss: 3.1289\n",
      "Epoch 21/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1254 - val_loss: 3.1313\n",
      "Epoch 22/300\n",
      "13519/13519 [==============================] - 14s 999us/step - loss: 3.1257 - val_loss: 3.1286\n",
      "Epoch 23/300\n",
      "13519/13519 [==============================] - 12s 860us/step - loss: 3.1236 - val_loss: 3.1264\n",
      "Epoch 24/300\n",
      "13519/13519 [==============================] - 12s 852us/step - loss: 3.1265 - val_loss: 3.1259\n",
      "Epoch 25/300\n",
      "13519/13519 [==============================] - 12s 924us/step - loss: 3.1231 - val_loss: 3.1248\n",
      "Epoch 26/300\n",
      "13519/13519 [==============================] - 12s 908us/step - loss: 3.1232 - val_loss: 3.1316\n",
      "Epoch 27/300\n",
      "13519/13519 [==============================] - 12s 872us/step - loss: 3.1249 - val_loss: 3.1255\n",
      "Epoch 28/300\n",
      "13519/13519 [==============================] - 13s 961us/step - loss: 3.1248 - val_loss: 3.1250\n",
      "Epoch 29/300\n",
      "13519/13519 [==============================] - 15s 1ms/step - loss: 3.1244 - val_loss: 3.1248\n",
      "Epoch 30/300\n",
      "13519/13519 [==============================] - 13s 964us/step - loss: 3.1227 - val_loss: 3.1236\n",
      "Epoch 31/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1231 - val_loss: 3.1251\n",
      "Epoch 32/300\n",
      "13519/13519 [==============================] - 13s 980us/step - loss: 3.1221 - val_loss: 3.1242\n",
      "Epoch 33/300\n",
      "13519/13519 [==============================] - 13s 985us/step - loss: 3.1234 - val_loss: 3.1259\n",
      "Epoch 34/300\n",
      "13519/13519 [==============================] - 13s 976us/step - loss: 3.1230 - val_loss: 3.1263\n",
      "Epoch 35/300\n",
      "13519/13519 [==============================] - 14s 999us/step - loss: 3.1249 - val_loss: 3.1251\n",
      "Epoch 36/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1221 - val_loss: 3.1232\n",
      "Epoch 37/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1212 - val_loss: 3.1227\n",
      "Epoch 38/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1222 - val_loss: 3.1236\n",
      "Epoch 39/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1225 - val_loss: 3.1234\n",
      "Epoch 40/300\n",
      "13519/13519 [==============================] - 15s 1ms/step - loss: 3.1223 - val_loss: 3.1224\n",
      "Epoch 41/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1219 - val_loss: 3.1259 ETA: 1s - loss - ETA: 0 - ETA: 0s - loss: 3.12\n",
      "Epoch 42/300\n",
      "13519/13519 [==============================] - 15s 1ms/step - loss: 3.1223 - val_loss: 3.1227\n",
      "Epoch 43/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1231 - val_loss: 3.1220\n",
      "Epoch 44/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1201 - val_loss: 3.1252\n",
      "Epoch 45/300\n",
      "13519/13519 [==============================] - 15s 1ms/step - loss: 3.1189 - val_loss: 3.1262\n",
      "Epoch 46/300\n",
      "13519/13519 [==============================] - 14s 999us/step - loss: 3.1211 - val_loss: 3.1454\n",
      "Epoch 47/300\n",
      "13519/13519 [==============================] - 16s 1ms/step - loss: 3.1206 - val_loss: 3.1226\n",
      "Epoch 48/300\n",
      "13519/13519 [==============================] - 19s 1ms/step - loss: 3.1200 - val_loss: 3.1208\n",
      "Epoch 49/300\n",
      "13519/13519 [==============================] - 16s 1ms/step - loss: 3.1197 - val_loss: 3.1241\n",
      "Epoch 50/300\n",
      "13519/13519 [==============================] - 15s 1ms/step - loss: 3.1210 - val_loss: 3.1218\n",
      "Epoch 51/300\n",
      "13519/13519 [==============================] - 13s 957us/step - loss: 3.1195 - val_loss: 3.1218\n",
      "Epoch 52/300\n",
      "13519/13519 [==============================] - 13s 984us/step - loss: 3.1197 - val_loss: 3.1205\n",
      "Epoch 53/300\n",
      "13519/13519 [==============================] - 13s 971us/step - loss: 3.1195 - val_loss: 3.1223\n",
      "Epoch 54/300\n",
      "13519/13519 [==============================] - 13s 975us/step - loss: 3.1211 - val_loss: 3.1239\n",
      "Epoch 55/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1195 - val_loss: 3.1232\n",
      "Epoch 56/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1211 - val_loss: 3.1231\n",
      "Epoch 57/300\n",
      "13519/13519 [==============================] - 15s 1ms/step - loss: 3.1192 - val_loss: 3.1205\n",
      "Epoch 58/300\n",
      "13519/13519 [==============================] - 13s 962us/step - loss: 3.1197 - val_loss: 3.1206\n",
      "Epoch 59/300\n",
      "13519/13519 [==============================] - 13s 981us/step - loss: 3.1194 - val_loss: 3.1206\n",
      "Epoch 60/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1174 - val_loss: 3.1228s: 3.\n",
      "Epoch 61/300\n",
      "13519/13519 [==============================] - 13s 939us/step - loss: 3.1232 - val_loss: 3.1208\n",
      "Epoch 62/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1206 - val_loss: 3.1210\n",
      "Epoch 63/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1178 - val_loss: 3.1205\n",
      "Epoch 64/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1174 - val_loss: 3.1225\n",
      "Epoch 65/300\n",
      "13519/13519 [==============================] - 14s 999us/step - loss: 3.1233 - val_loss: 3.1201\n",
      "Epoch 66/300\n",
      "13519/13519 [==============================] - 13s 948us/step - loss: 3.1174 - val_loss: 3.1212\n",
      "Epoch 67/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1188 - val_loss: 3.1203\n",
      "Epoch 68/300\n",
      "13519/13519 [==============================] - 16s 1ms/step - loss: 3.1185 - val_loss: 3.1207\n",
      "Epoch 69/300\n",
      "13519/13519 [==============================] - 16s 1ms/step - loss: 3.1183 - val_loss: 3.1210\n",
      "Epoch 70/300\n",
      "13519/13519 [==============================] - 11s 833us/step - loss: 3.1196 - val_loss: 3.1236\n",
      "Epoch 71/300\n",
      "13519/13519 [==============================] - 11s 807us/step - loss: 3.1195 - val_loss: 3.1203\n",
      "Epoch 72/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1203 - val_loss: 3.1233\n",
      "Epoch 73/300\n",
      "13519/13519 [==============================] - 13s 994us/step - loss: 3.1196 - val_loss: 3.1206\n",
      "Epoch 74/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1175 - val_loss: 3.1225\n",
      "Epoch 75/300\n",
      "13519/13519 [==============================] - 13s 956us/step - loss: 3.1182 - val_loss: 3.1204\n",
      "Epoch 76/300\n",
      "13519/13519 [==============================] - 13s 974us/step - loss: 3.1179 - val_loss: 3.1252\n",
      "Epoch 77/300\n",
      "13519/13519 [==============================] - 17s 1ms/step - loss: 3.1172 - val_loss: 3.1199\n",
      "Epoch 78/300\n",
      "13519/13519 [==============================] - 15s 1ms/step - loss: 3.1200 - val_loss: 3.1208\n",
      "Epoch 79/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1185 - val_loss: 3.1255\n",
      "Epoch 80/300\n",
      "13519/13519 [==============================] - 12s 886us/step - loss: 3.1166 - val_loss: 3.1247\n",
      "Epoch 81/300\n",
      "13519/13519 [==============================] - 11s 831us/step - loss: 3.1189 - val_loss: 3.1191\n",
      "Epoch 82/300\n",
      "13519/13519 [==============================] - 12s 852us/step - loss: 3.1191 - val_loss: 3.1211\n",
      "Epoch 83/300\n",
      "13519/13519 [==============================] - 13s 996us/step - loss: 3.1176 - val_loss: 3.1203\n",
      "Epoch 84/300\n",
      "13519/13519 [==============================] - 12s 869us/step - loss: 3.1162 - val_loss: 3.1240\n",
      "Epoch 85/300\n",
      "13519/13519 [==============================] - 12s 922us/step - loss: 3.1179 - val_loss: 3.1195\n",
      "Epoch 86/300\n",
      "13519/13519 [==============================] - 11s 803us/step - loss: 3.1164 - val_loss: 3.1278\n",
      "Epoch 87/300\n",
      "13519/13519 [==============================] - 14s 1000us/step - loss: 3.1205 - val_loss: 3.1221\n",
      "Epoch 88/300\n",
      "13519/13519 [==============================] - 13s 965us/step - loss: 3.1192 - val_loss: 3.1198\n",
      "Epoch 89/300\n",
      "13519/13519 [==============================] - 13s 961us/step - loss: 3.1166 - val_loss: 3.1198\n",
      "Epoch 90/300\n",
      "13519/13519 [==============================] - 13s 962us/step - loss: 3.1173 - val_loss: 3.1519\n",
      "Epoch 91/300\n",
      "13519/13519 [==============================] - 13s 955us/step - loss: 3.1200 - val_loss: 3.1202\n",
      "Epoch 92/300\n",
      "13519/13519 [==============================] - 13s 976us/step - loss: 3.1171 - val_loss: 3.1210\n",
      "Epoch 93/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1182 - val_loss: 3.1206\n",
      "Epoch 94/300\n",
      "13519/13519 [==============================] - 15s 1ms/step - loss: 3.1180 - val_loss: 3.1197\n",
      "Epoch 95/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1193 - val_loss: 3.1189\n",
      "Epoch 96/300\n",
      "13519/13519 [==============================] - 13s 983us/step - loss: 3.1178 - val_loss: 3.1191\n",
      "Epoch 97/300\n",
      "13519/13519 [==============================] - 11s 803us/step - loss: 3.1179 - val_loss: 3.1216\n",
      "Epoch 98/300\n",
      "13519/13519 [==============================] - 11s 818us/step - loss: 3.1166 - val_loss: 3.1195\n",
      "Epoch 99/300\n",
      "13519/13519 [==============================] - 12s 855us/step - loss: 3.1169 - val_loss: 3.1190\n",
      "Epoch 100/300\n",
      "13519/13519 [==============================] - 12s 898us/step - loss: 3.1174 - val_loss: 3.1248\n",
      "Epoch 101/300\n",
      "13519/13519 [==============================] - 12s 875us/step - loss: 3.1171 - val_loss: 3.1223\n",
      "Epoch 102/300\n",
      "13519/13519 [==============================] - 13s 962us/step - loss: 3.1166 - val_loss: 3.1187\n",
      "Epoch 103/300\n",
      "13519/13519 [==============================] - 12s 874us/step - loss: 3.1151 - val_loss: 3.1185\n",
      "Epoch 104/300\n",
      "13519/13519 [==============================] - 11s 827us/step - loss: 3.1164 - val_loss: 3.1237\n",
      "Epoch 105/300\n",
      "13519/13519 [==============================] - 11s 811us/step - loss: 3.1178 - val_loss: 3.1204\n",
      "Epoch 106/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1169 - val_loss: 3.1201\n",
      "Epoch 107/300\n",
      "13519/13519 [==============================] - 13s 940us/step - loss: 3.1178 - val_loss: 3.1184\n",
      "Epoch 108/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1158 - val_loss: 3.1231\n",
      "Epoch 109/300\n",
      "13519/13519 [==============================] - 11s 790us/step - loss: 3.1155 - val_loss: 3.1189\n",
      "Epoch 110/300\n",
      "13519/13519 [==============================] - 11s 781us/step - loss: 3.1174 - val_loss: 3.1235\n",
      "Epoch 111/300\n",
      "13519/13519 [==============================] - 12s 907us/step - loss: 3.1214 - val_loss: 3.1176\n",
      "Epoch 112/300\n",
      "13519/13519 [==============================] - 13s 930us/step - loss: 3.1154 - val_loss: 3.1206\n",
      "Epoch 113/300\n",
      "13519/13519 [==============================] - 12s 894us/step - loss: 3.1165 - val_loss: 3.1204\n",
      "Epoch 114/300\n",
      "13519/13519 [==============================] - 12s 893us/step - loss: 3.1156 - val_loss: 3.1203\n",
      "Epoch 115/300\n",
      "13519/13519 [==============================] - 13s 970us/step - loss: 3.1179 - val_loss: 3.1207\n",
      "Epoch 116/300\n",
      "13519/13519 [==============================] - 13s 947us/step - loss: 3.1174 - val_loss: 3.1221\n",
      "Epoch 117/300\n",
      "13519/13519 [==============================] - 15s 1ms/step - loss: 3.1154 - val_loss: 3.1291\n",
      "Epoch 118/300\n",
      "13519/13519 [==============================] - 11s 824us/step - loss: 3.1169 - val_loss: 3.1185\n",
      "Epoch 119/300\n",
      "13519/13519 [==============================] - 12s 903us/step - loss: 3.1168 - val_loss: 3.1209\n",
      "Epoch 120/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1158 - val_loss: 3.1190\n",
      "Epoch 121/300\n",
      "13519/13519 [==============================] - 13s 964us/step - loss: 3.1157 - val_loss: 3.1200\n",
      "Epoch 122/300\n",
      "13519/13519 [==============================] - 13s 968us/step - loss: 3.1180 - val_loss: 3.1196\n",
      "Epoch 123/300\n",
      "13519/13519 [==============================] - 13s 974us/step - loss: 3.1151 - val_loss: 3.1221\n",
      "Epoch 124/300\n",
      "13519/13519 [==============================] - 13s 960us/step - loss: 3.1198 - val_loss: 3.1193\n",
      "Epoch 125/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1177 - val_loss: 3.1194\n",
      "Epoch 126/300\n",
      "13519/13519 [==============================] - 14s 999us/step - loss: 3.1168 - val_loss: 3.1199\n",
      "Epoch 127/300\n",
      "13519/13519 [==============================] - 13s 967us/step - loss: 3.1163 - val_loss: 3.1189\n",
      "Epoch 128/300\n",
      "13519/13519 [==============================] - 13s 966us/step - loss: 3.1246 - val_loss: 3.1214\n",
      "Epoch 129/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1176 - val_loss: 3.1187\n",
      "Epoch 130/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1132 - val_loss: 3.1181\n",
      "Epoch 131/300\n",
      "13519/13519 [==============================] - 14s 1ms/step - loss: 3.1160 - val_loss: 3.1203\n",
      "Epoch 00131: early stopping\n"
     ]
    }
   ],
   "source": [
    "np.random.seed(10)\n",
    "tf.random.set_seed(20)\n",
    "random.seed(30)    \n",
    "\n",
    "epochs = 300\n",
    "\n",
    "# very simple NN\n",
    "model = Sequential()\n",
    "model.add(Dense(32, input_dim=X_train.shape[1], activation='relu'))\n",
    "model.add(Dense(30, activation='softmax'))\n",
    "\n",
    "adam = adam_v2.Adam(learning_rate=0.0005) # default: 0.001\n",
    "model.compile(loss='categorical_crossentropy', optimizer=adam)\n",
    "\n",
    "es = EarlyStopping(monitor='val_loss', patience=20, verbose=1)\n",
    "h = model.fit(\n",
    "    X_train,\n",
    "    to_categorical(y_train, num_classes=30),\n",
    "    validation_data=(X_val, to_categorical(y_val, num_classes=30)),\n",
    "    epochs=epochs,\n",
    "    batch_size=32,\n",
    "    callbacks=[es]\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "f106e448",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(h.history['loss'])\n",
    "plt.plot(h.history['val_loss'])\n",
    "plt.legend(['Train', 'Val'])\n",
    "plt.grid();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "865151da",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Log loss (train): 3.1138222741611132\n",
      "Log loss (val): 3.1192589680188854\n",
      "RPS (train): 3.662961944065668\n",
      "RPS (val): 3.678533428175507\n",
      "CPU times: user 10.1 s, sys: 1.99 s, total: 12.1 s\n",
      "Wall time: 9.14 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "# predictions\n",
    "probs_train = model.predict(X_train)\n",
    "probs_val = model.predict(X_val)\n",
    "\n",
    "log_train = log_loss(y_train, probs_train)\n",
    "log_val = log_loss(y_val, probs_val)\n",
    "print('Log loss (train):', log_train)\n",
    "print('Log loss (val):', log_val)\n",
    "\n",
    "rps_train = scoring_function(y_train + 1, probs_train)\n",
    "rps_val = scoring_function(y_val + 1, probs_val)\n",
    "print('RPS (train):', rps_train)\n",
    "print('RPS (val):', rps_val)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6fe4d5ae",
   "metadata": {},
   "source": [
    "### Save predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "80d3058c",
   "metadata": {},
   "outputs": [],
   "source": [
    "pd.DataFrame(probs_val).to_csv('3.1-probs-nn-one-layer-32.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "191c923b",
   "metadata": {},
   "source": [
    "## Submission"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "db10f692",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>sold_quantity_17</th>\n",
       "      <th>sold_quantity_18</th>\n",
       "      <th>sold_quantity_19</th>\n",
       "      <th>sold_quantity_20</th>\n",
       "      <th>sold_quantity_21</th>\n",
       "      <th>sold_quantity_25</th>\n",
       "      <th>sold_quantity_26</th>\n",
       "      <th>sold_quantity_27</th>\n",
       "      <th>sold_quantity_28</th>\n",
       "      <th>sold_quantity_29</th>\n",
       "      <th>sold_quantity_30</th>\n",
       "      <th>minutes_active_29</th>\n",
       "      <th>minutes_active_30</th>\n",
       "      <th>roll_mean_sold_30_11</th>\n",
       "      <th>roll_mean_sold_29_10</th>\n",
       "      <th>roll_mean_sold_28_9</th>\n",
       "      <th>roll_mean_sold_27_8</th>\n",
       "      <th>roll_mean_sold_26_7</th>\n",
       "      <th>roll_mean_sold_25_6</th>\n",
       "      <th>roll_mean_sold_24_5</th>\n",
       "      <th>roll_mean_minutes_30_11</th>\n",
       "      <th>roll_mean_minutes_29_10</th>\n",
       "      <th>roll_mean_minutes_26_7</th>\n",
       "      <th>roll_mean_minutes_25_6</th>\n",
       "      <th>roll_mean_minutes_24_5</th>\n",
       "      <th>roll_mean_price_30_11</th>\n",
       "      <th>roll_mean_price_29_10</th>\n",
       "      <th>roll_mean_price_27_8</th>\n",
       "      <th>roll_mean_price_25_6</th>\n",
       "      <th>roll_mean_price_24_5</th>\n",
       "      <th>roll_sum_zero_sold_30_11</th>\n",
       "      <th>roll_sum_zero_sold_24_5</th>\n",
       "      <th>roll_sum_shipping_logistic_type_fulfillment_30_11</th>\n",
       "      <th>roll_sum_shipping_logistic_type_cross_docking_30_11</th>\n",
       "      <th>roll_sum_shipping_logistic_type_fulfillment_28_9</th>\n",
       "      <th>roll_sum_shipping_logistic_type_fulfillment_26_7</th>\n",
       "      <th>roll_sum_shipping_logistic_type_fulfillment_25_6</th>\n",
       "      <th>roll_sum_shipping_logistic_type_fulfillment_24_5</th>\n",
       "      <th>roll_sum_shipping_payment_free_shipping_28_9</th>\n",
       "      <th>target_stock</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sku</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",
       "      <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",
       "      <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>464801</th>\n",
       "      <td>-0.096217</td>\n",
       "      <td>-0.090172</td>\n",
       "      <td>-0.099056</td>\n",
       "      <td>-0.10158</td>\n",
       "      <td>0.097635</td>\n",
       "      <td>-0.084454</td>\n",
       "      <td>-0.079735</td>\n",
       "      <td>-0.102219</td>\n",
       "      <td>-0.094906</td>\n",
       "      <td>-0.118098</td>\n",
       "      <td>-0.115690</td>\n",
       "      <td>0.530162</td>\n",
       "      <td>0.533972</td>\n",
       "      <td>-0.095625</td>\n",
       "      <td>-0.094022</td>\n",
       "      <td>-0.087551</td>\n",
       "      <td>-0.087259</td>\n",
       "      <td>-0.087144</td>\n",
       "      <td>-0.087088</td>\n",
       "      <td>-0.086720</td>\n",
       "      <td>0.737185</td>\n",
       "      <td>0.744720</td>\n",
       "      <td>0.763959</td>\n",
       "      <td>0.769021</td>\n",
       "      <td>0.774894</td>\n",
       "      <td>0.000237</td>\n",
       "      <td>0.000191</td>\n",
       "      <td>-0.004176</td>\n",
       "      <td>-0.00350</td>\n",
       "      <td>-0.003556</td>\n",
       "      <td>0.527432</td>\n",
       "      <td>0.314442</td>\n",
       "      <td>0.479567</td>\n",
       "      <td>-0.297576</td>\n",
       "      <td>0.494024</td>\n",
       "      <td>0.50752</td>\n",
       "      <td>0.514066</td>\n",
       "      <td>0.520863</td>\n",
       "      <td>1.490386</td>\n",
       "      <td>-0.124437</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>645793</th>\n",
       "      <td>-0.000423</td>\n",
       "      <td>0.074500</td>\n",
       "      <td>-0.099056</td>\n",
       "      <td>-0.10158</td>\n",
       "      <td>0.018155</td>\n",
       "      <td>-0.084454</td>\n",
       "      <td>-0.079735</td>\n",
       "      <td>-0.102219</td>\n",
       "      <td>0.009552</td>\n",
       "      <td>-0.118098</td>\n",
       "      <td>-0.115690</td>\n",
       "      <td>0.530162</td>\n",
       "      <td>0.533972</td>\n",
       "      <td>-0.077280</td>\n",
       "      <td>-0.075858</td>\n",
       "      <td>-0.069648</td>\n",
       "      <td>-0.063206</td>\n",
       "      <td>-0.062798</td>\n",
       "      <td>-0.062530</td>\n",
       "      <td>-0.062069</td>\n",
       "      <td>0.737185</td>\n",
       "      <td>0.744720</td>\n",
       "      <td>0.763959</td>\n",
       "      <td>0.769021</td>\n",
       "      <td>0.774894</td>\n",
       "      <td>0.000237</td>\n",
       "      <td>0.000191</td>\n",
       "      <td>-0.004176</td>\n",
       "      <td>-0.00350</td>\n",
       "      <td>-0.003556</td>\n",
       "      <td>-0.073491</td>\n",
       "      <td>-0.266020</td>\n",
       "      <td>0.479567</td>\n",
       "      <td>-0.297576</td>\n",
       "      <td>0.494024</td>\n",
       "      <td>0.50752</td>\n",
       "      <td>0.514066</td>\n",
       "      <td>0.520863</td>\n",
       "      <td>1.490386</td>\n",
       "      <td>-0.116708</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>99516</th>\n",
       "      <td>-0.096217</td>\n",
       "      <td>-0.090172</td>\n",
       "      <td>-0.099056</td>\n",
       "      <td>-0.10158</td>\n",
       "      <td>-0.061326</td>\n",
       "      <td>0.064437</td>\n",
       "      <td>-0.079735</td>\n",
       "      <td>0.411154</td>\n",
       "      <td>0.427382</td>\n",
       "      <td>0.277992</td>\n",
       "      <td>0.364686</td>\n",
       "      <td>0.530162</td>\n",
       "      <td>0.533972</td>\n",
       "      <td>0.045019</td>\n",
       "      <td>0.027074</td>\n",
       "      <td>0.013903</td>\n",
       "      <td>0.002940</td>\n",
       "      <td>-0.008019</td>\n",
       "      <td>0.011142</td>\n",
       "      <td>0.005720</td>\n",
       "      <td>0.117517</td>\n",
       "      <td>0.128402</td>\n",
       "      <td>0.160092</td>\n",
       "      <td>0.169554</td>\n",
       "      <td>0.179085</td>\n",
       "      <td>-0.004385</td>\n",
       "      <td>-0.004378</td>\n",
       "      <td>-0.004405</td>\n",
       "      <td>-0.00362</td>\n",
       "      <td>-0.003677</td>\n",
       "      <td>-0.474106</td>\n",
       "      <td>-0.652995</td>\n",
       "      <td>-0.792469</td>\n",
       "      <td>-0.297576</td>\n",
       "      <td>-0.744423</td>\n",
       "      <td>-0.69828</td>\n",
       "      <td>-0.676118</td>\n",
       "      <td>-0.654889</td>\n",
       "      <td>-0.686343</td>\n",
       "      <td>-0.085792</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>538100</th>\n",
       "      <td>-0.000423</td>\n",
       "      <td>-0.090172</td>\n",
       "      <td>-0.099056</td>\n",
       "      <td>-0.10158</td>\n",
       "      <td>-0.061326</td>\n",
       "      <td>-0.010009</td>\n",
       "      <td>-0.079735</td>\n",
       "      <td>0.000456</td>\n",
       "      <td>0.114009</td>\n",
       "      <td>0.277992</td>\n",
       "      <td>0.268610</td>\n",
       "      <td>0.530162</td>\n",
       "      <td>0.533972</td>\n",
       "      <td>-0.016131</td>\n",
       "      <td>-0.021365</td>\n",
       "      <td>-0.039808</td>\n",
       "      <td>-0.045166</td>\n",
       "      <td>-0.050625</td>\n",
       "      <td>-0.044112</td>\n",
       "      <td>-0.049744</td>\n",
       "      <td>0.737185</td>\n",
       "      <td>0.744720</td>\n",
       "      <td>0.763959</td>\n",
       "      <td>0.769021</td>\n",
       "      <td>0.774894</td>\n",
       "      <td>-0.004385</td>\n",
       "      <td>-0.004378</td>\n",
       "      <td>-0.004405</td>\n",
       "      <td>-0.00362</td>\n",
       "      <td>-0.003677</td>\n",
       "      <td>-0.874720</td>\n",
       "      <td>-0.459507</td>\n",
       "      <td>0.479567</td>\n",
       "      <td>-0.297576</td>\n",
       "      <td>0.494024</td>\n",
       "      <td>0.50752</td>\n",
       "      <td>0.514066</td>\n",
       "      <td>0.520863</td>\n",
       "      <td>-0.686343</td>\n",
       "      <td>-0.085792</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>557191</th>\n",
       "      <td>-0.096217</td>\n",
       "      <td>-0.090172</td>\n",
       "      <td>-0.099056</td>\n",
       "      <td>-0.10158</td>\n",
       "      <td>-0.061326</td>\n",
       "      <td>-0.084454</td>\n",
       "      <td>-0.079735</td>\n",
       "      <td>-0.102219</td>\n",
       "      <td>-0.094906</td>\n",
       "      <td>-0.118098</td>\n",
       "      <td>0.076460</td>\n",
       "      <td>0.530162</td>\n",
       "      <td>0.533972</td>\n",
       "      <td>-0.107855</td>\n",
       "      <td>-0.118242</td>\n",
       "      <td>-0.117391</td>\n",
       "      <td>-0.117325</td>\n",
       "      <td>-0.105404</td>\n",
       "      <td>-0.105506</td>\n",
       "      <td>-0.105208</td>\n",
       "      <td>-1.222632</td>\n",
       "      <td>-1.352601</td>\n",
       "      <td>-1.631688</td>\n",
       "      <td>-1.609169</td>\n",
       "      <td>-1.588786</td>\n",
       "      <td>-0.004385</td>\n",
       "      <td>-0.004378</td>\n",
       "      <td>-0.004405</td>\n",
       "      <td>-0.00362</td>\n",
       "      <td>-0.003677</td>\n",
       "      <td>0.928047</td>\n",
       "      <td>0.894904</td>\n",
       "      <td>0.479567</td>\n",
       "      <td>-0.297576</td>\n",
       "      <td>0.494024</td>\n",
       "      <td>0.50752</td>\n",
       "      <td>0.514066</td>\n",
       "      <td>0.520863</td>\n",
       "      <td>-0.686343</td>\n",
       "      <td>-0.070334</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>129187</th>\n",
       "      <td>-0.096217</td>\n",
       "      <td>-0.090172</td>\n",
       "      <td>-0.099056</td>\n",
       "      <td>-0.10158</td>\n",
       "      <td>-0.061326</td>\n",
       "      <td>-0.084454</td>\n",
       "      <td>-0.079735</td>\n",
       "      <td>-0.102219</td>\n",
       "      <td>-0.094906</td>\n",
       "      <td>-0.118098</td>\n",
       "      <td>-0.115690</td>\n",
       "      <td>-1.943069</td>\n",
       "      <td>-1.502754</td>\n",
       "      <td>-0.120085</td>\n",
       "      <td>-0.118242</td>\n",
       "      <td>-0.117391</td>\n",
       "      <td>-0.117325</td>\n",
       "      <td>-0.117577</td>\n",
       "      <td>-0.117785</td>\n",
       "      <td>-0.117533</td>\n",
       "      <td>-2.213161</td>\n",
       "      <td>-2.217213</td>\n",
       "      <td>-2.138135</td>\n",
       "      <td>-2.111925</td>\n",
       "      <td>-2.088475</td>\n",
       "      <td>-0.004385</td>\n",
       "      <td>-0.004378</td>\n",
       "      <td>-0.004405</td>\n",
       "      <td>-0.00362</td>\n",
       "      <td>-0.003677</td>\n",
       "      <td>-2.677488</td>\n",
       "      <td>-2.781355</td>\n",
       "      <td>-2.700524</td>\n",
       "      <td>-0.297576</td>\n",
       "      <td>-2.602093</td>\n",
       "      <td>-2.50698</td>\n",
       "      <td>-2.461394</td>\n",
       "      <td>-2.418515</td>\n",
       "      <td>-0.686343</td>\n",
       "      <td>-0.085792</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6707</th>\n",
       "      <td>-0.096217</td>\n",
       "      <td>-0.090172</td>\n",
       "      <td>-0.099056</td>\n",
       "      <td>-0.10158</td>\n",
       "      <td>-0.061326</td>\n",
       "      <td>-0.084454</td>\n",
       "      <td>-0.079735</td>\n",
       "      <td>-0.102219</td>\n",
       "      <td>-0.094906</td>\n",
       "      <td>-0.118098</td>\n",
       "      <td>-0.115690</td>\n",
       "      <td>-1.943069</td>\n",
       "      <td>-1.505581</td>\n",
       "      <td>-0.120085</td>\n",
       "      <td>-0.118242</td>\n",
       "      <td>-0.117391</td>\n",
       "      <td>-0.117325</td>\n",
       "      <td>-0.117577</td>\n",
       "      <td>-0.117785</td>\n",
       "      <td>-0.117533</td>\n",
       "      <td>-2.213329</td>\n",
       "      <td>-2.217213</td>\n",
       "      <td>-2.138135</td>\n",
       "      <td>-2.111925</td>\n",
       "      <td>-2.088475</td>\n",
       "      <td>-0.004385</td>\n",
       "      <td>-0.004378</td>\n",
       "      <td>-0.004405</td>\n",
       "      <td>-0.00362</td>\n",
       "      <td>-0.003677</td>\n",
       "      <td>-2.677488</td>\n",
       "      <td>-2.781355</td>\n",
       "      <td>-2.700524</td>\n",
       "      <td>-0.060230</td>\n",
       "      <td>-2.602093</td>\n",
       "      <td>-2.50698</td>\n",
       "      <td>-2.461394</td>\n",
       "      <td>-2.418515</td>\n",
       "      <td>-0.686343</td>\n",
       "      <td>0.084246</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>170355</th>\n",
       "      <td>-0.096217</td>\n",
       "      <td>-0.090172</td>\n",
       "      <td>-0.099056</td>\n",
       "      <td>-0.10158</td>\n",
       "      <td>-0.061326</td>\n",
       "      <td>-0.084454</td>\n",
       "      <td>-0.079735</td>\n",
       "      <td>-0.102219</td>\n",
       "      <td>-0.094906</td>\n",
       "      <td>-0.118098</td>\n",
       "      <td>-0.115690</td>\n",
       "      <td>-1.943069</td>\n",
       "      <td>-1.967433</td>\n",
       "      <td>-0.120085</td>\n",
       "      <td>-0.118242</td>\n",
       "      <td>-0.117391</td>\n",
       "      <td>-0.117325</td>\n",
       "      <td>-0.117577</td>\n",
       "      <td>-0.117785</td>\n",
       "      <td>-0.117533</td>\n",
       "      <td>-2.240817</td>\n",
       "      <td>-2.217213</td>\n",
       "      <td>-2.138135</td>\n",
       "      <td>-2.111925</td>\n",
       "      <td>-2.088475</td>\n",
       "      <td>-0.004385</td>\n",
       "      <td>-0.004378</td>\n",
       "      <td>-0.004405</td>\n",
       "      <td>-0.00362</td>\n",
       "      <td>-0.003677</td>\n",
       "      <td>-2.677488</td>\n",
       "      <td>-2.781355</td>\n",
       "      <td>-2.700524</td>\n",
       "      <td>-0.297576</td>\n",
       "      <td>-2.602093</td>\n",
       "      <td>-2.50698</td>\n",
       "      <td>-2.461394</td>\n",
       "      <td>-2.418515</td>\n",
       "      <td>-0.686343</td>\n",
       "      <td>-0.124437</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>246568</th>\n",
       "      <td>-0.096217</td>\n",
       "      <td>-0.090172</td>\n",
       "      <td>-0.099056</td>\n",
       "      <td>-0.10158</td>\n",
       "      <td>-0.061326</td>\n",
       "      <td>-0.084454</td>\n",
       "      <td>-0.079735</td>\n",
       "      <td>-0.102219</td>\n",
       "      <td>-0.094906</td>\n",
       "      <td>-0.118098</td>\n",
       "      <td>-0.115690</td>\n",
       "      <td>-1.943069</td>\n",
       "      <td>-1.732601</td>\n",
       "      <td>-0.120085</td>\n",
       "      <td>-0.118242</td>\n",
       "      <td>-0.117391</td>\n",
       "      <td>-0.117325</td>\n",
       "      <td>-0.117577</td>\n",
       "      <td>-0.117785</td>\n",
       "      <td>-0.117533</td>\n",
       "      <td>-2.226841</td>\n",
       "      <td>-2.217213</td>\n",
       "      <td>-2.138135</td>\n",
       "      <td>-2.111925</td>\n",
       "      <td>-2.088475</td>\n",
       "      <td>-0.004385</td>\n",
       "      <td>-0.004378</td>\n",
       "      <td>-0.004405</td>\n",
       "      <td>-0.00362</td>\n",
       "      <td>-0.003677</td>\n",
       "      <td>-2.677488</td>\n",
       "      <td>-2.781355</td>\n",
       "      <td>-2.541519</td>\n",
       "      <td>-0.297576</td>\n",
       "      <td>-2.602093</td>\n",
       "      <td>-2.50698</td>\n",
       "      <td>-2.461394</td>\n",
       "      <td>-2.418515</td>\n",
       "      <td>-0.686343</td>\n",
       "      <td>-0.132166</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49718</th>\n",
       "      <td>-0.096217</td>\n",
       "      <td>-0.090172</td>\n",
       "      <td>-0.099056</td>\n",
       "      <td>-0.10158</td>\n",
       "      <td>-0.061326</td>\n",
       "      <td>-0.084454</td>\n",
       "      <td>-0.079735</td>\n",
       "      <td>-0.102219</td>\n",
       "      <td>-0.094906</td>\n",
       "      <td>-0.118098</td>\n",
       "      <td>-0.115690</td>\n",
       "      <td>-1.943069</td>\n",
       "      <td>-1.814924</td>\n",
       "      <td>-0.120085</td>\n",
       "      <td>-0.118242</td>\n",
       "      <td>-0.117391</td>\n",
       "      <td>-0.117325</td>\n",
       "      <td>-0.117577</td>\n",
       "      <td>-0.117785</td>\n",
       "      <td>-0.117533</td>\n",
       "      <td>-2.231740</td>\n",
       "      <td>-2.217213</td>\n",
       "      <td>-2.138135</td>\n",
       "      <td>-2.111925</td>\n",
       "      <td>-2.088475</td>\n",
       "      <td>-0.004385</td>\n",
       "      <td>-0.004378</td>\n",
       "      <td>-0.004405</td>\n",
       "      <td>-0.00362</td>\n",
       "      <td>-0.003677</td>\n",
       "      <td>-2.677488</td>\n",
       "      <td>-2.781355</td>\n",
       "      <td>-2.700524</td>\n",
       "      <td>-0.297576</td>\n",
       "      <td>-2.602093</td>\n",
       "      <td>-2.50698</td>\n",
       "      <td>-2.461394</td>\n",
       "      <td>-2.418515</td>\n",
       "      <td>-0.686343</td>\n",
       "      <td>0.006956</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>551472 rows × 40 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        sold_quantity_17  sold_quantity_18  sold_quantity_19  \\\n",
       "sku                                                            \n",
       "464801         -0.096217         -0.090172         -0.099056   \n",
       "645793         -0.000423          0.074500         -0.099056   \n",
       "99516          -0.096217         -0.090172         -0.099056   \n",
       "538100         -0.000423         -0.090172         -0.099056   \n",
       "557191         -0.096217         -0.090172         -0.099056   \n",
       "...                  ...               ...               ...   \n",
       "129187         -0.096217         -0.090172         -0.099056   \n",
       "6707           -0.096217         -0.090172         -0.099056   \n",
       "170355         -0.096217         -0.090172         -0.099056   \n",
       "246568         -0.096217         -0.090172         -0.099056   \n",
       "49718          -0.096217         -0.090172         -0.099056   \n",
       "\n",
       "        sold_quantity_20  sold_quantity_21  sold_quantity_25  \\\n",
       "sku                                                            \n",
       "464801          -0.10158          0.097635         -0.084454   \n",
       "645793          -0.10158          0.018155         -0.084454   \n",
       "99516           -0.10158         -0.061326          0.064437   \n",
       "538100          -0.10158         -0.061326         -0.010009   \n",
       "557191          -0.10158         -0.061326         -0.084454   \n",
       "...                  ...               ...               ...   \n",
       "129187          -0.10158         -0.061326         -0.084454   \n",
       "6707            -0.10158         -0.061326         -0.084454   \n",
       "170355          -0.10158         -0.061326         -0.084454   \n",
       "246568          -0.10158         -0.061326         -0.084454   \n",
       "49718           -0.10158         -0.061326         -0.084454   \n",
       "\n",
       "        sold_quantity_26  sold_quantity_27  sold_quantity_28  \\\n",
       "sku                                                            \n",
       "464801         -0.079735         -0.102219         -0.094906   \n",
       "645793         -0.079735         -0.102219          0.009552   \n",
       "99516          -0.079735          0.411154          0.427382   \n",
       "538100         -0.079735          0.000456          0.114009   \n",
       "557191         -0.079735         -0.102219         -0.094906   \n",
       "...                  ...               ...               ...   \n",
       "129187         -0.079735         -0.102219         -0.094906   \n",
       "6707           -0.079735         -0.102219         -0.094906   \n",
       "170355         -0.079735         -0.102219         -0.094906   \n",
       "246568         -0.079735         -0.102219         -0.094906   \n",
       "49718          -0.079735         -0.102219         -0.094906   \n",
       "\n",
       "        sold_quantity_29  sold_quantity_30  minutes_active_29  \\\n",
       "sku                                                             \n",
       "464801         -0.118098         -0.115690           0.530162   \n",
       "645793         -0.118098         -0.115690           0.530162   \n",
       "99516           0.277992          0.364686           0.530162   \n",
       "538100          0.277992          0.268610           0.530162   \n",
       "557191         -0.118098          0.076460           0.530162   \n",
       "...                  ...               ...                ...   \n",
       "129187         -0.118098         -0.115690          -1.943069   \n",
       "6707           -0.118098         -0.115690          -1.943069   \n",
       "170355         -0.118098         -0.115690          -1.943069   \n",
       "246568         -0.118098         -0.115690          -1.943069   \n",
       "49718          -0.118098         -0.115690          -1.943069   \n",
       "\n",
       "        minutes_active_30  roll_mean_sold_30_11  roll_mean_sold_29_10  \\\n",
       "sku                                                                     \n",
       "464801           0.533972             -0.095625             -0.094022   \n",
       "645793           0.533972             -0.077280             -0.075858   \n",
       "99516            0.533972              0.045019              0.027074   \n",
       "538100           0.533972             -0.016131             -0.021365   \n",
       "557191           0.533972             -0.107855             -0.118242   \n",
       "...                   ...                   ...                   ...   \n",
       "129187          -1.502754             -0.120085             -0.118242   \n",
       "6707            -1.505581             -0.120085             -0.118242   \n",
       "170355          -1.967433             -0.120085             -0.118242   \n",
       "246568          -1.732601             -0.120085             -0.118242   \n",
       "49718           -1.814924             -0.120085             -0.118242   \n",
       "\n",
       "        roll_mean_sold_28_9  roll_mean_sold_27_8  roll_mean_sold_26_7  \\\n",
       "sku                                                                     \n",
       "464801            -0.087551            -0.087259            -0.087144   \n",
       "645793            -0.069648            -0.063206            -0.062798   \n",
       "99516              0.013903             0.002940            -0.008019   \n",
       "538100            -0.039808            -0.045166            -0.050625   \n",
       "557191            -0.117391            -0.117325            -0.105404   \n",
       "...                     ...                  ...                  ...   \n",
       "129187            -0.117391            -0.117325            -0.117577   \n",
       "6707              -0.117391            -0.117325            -0.117577   \n",
       "170355            -0.117391            -0.117325            -0.117577   \n",
       "246568            -0.117391            -0.117325            -0.117577   \n",
       "49718             -0.117391            -0.117325            -0.117577   \n",
       "\n",
       "        roll_mean_sold_25_6  roll_mean_sold_24_5  roll_mean_minutes_30_11  \\\n",
       "sku                                                                         \n",
       "464801            -0.087088            -0.086720                 0.737185   \n",
       "645793            -0.062530            -0.062069                 0.737185   \n",
       "99516              0.011142             0.005720                 0.117517   \n",
       "538100            -0.044112            -0.049744                 0.737185   \n",
       "557191            -0.105506            -0.105208                -1.222632   \n",
       "...                     ...                  ...                      ...   \n",
       "129187            -0.117785            -0.117533                -2.213161   \n",
       "6707              -0.117785            -0.117533                -2.213329   \n",
       "170355            -0.117785            -0.117533                -2.240817   \n",
       "246568            -0.117785            -0.117533                -2.226841   \n",
       "49718             -0.117785            -0.117533                -2.231740   \n",
       "\n",
       "        roll_mean_minutes_29_10  roll_mean_minutes_26_7  \\\n",
       "sku                                                       \n",
       "464801                 0.744720                0.763959   \n",
       "645793                 0.744720                0.763959   \n",
       "99516                  0.128402                0.160092   \n",
       "538100                 0.744720                0.763959   \n",
       "557191                -1.352601               -1.631688   \n",
       "...                         ...                     ...   \n",
       "129187                -2.217213               -2.138135   \n",
       "6707                  -2.217213               -2.138135   \n",
       "170355                -2.217213               -2.138135   \n",
       "246568                -2.217213               -2.138135   \n",
       "49718                 -2.217213               -2.138135   \n",
       "\n",
       "        roll_mean_minutes_25_6  roll_mean_minutes_24_5  roll_mean_price_30_11  \\\n",
       "sku                                                                             \n",
       "464801                0.769021                0.774894               0.000237   \n",
       "645793                0.769021                0.774894               0.000237   \n",
       "99516                 0.169554                0.179085              -0.004385   \n",
       "538100                0.769021                0.774894              -0.004385   \n",
       "557191               -1.609169               -1.588786              -0.004385   \n",
       "...                        ...                     ...                    ...   \n",
       "129187               -2.111925               -2.088475              -0.004385   \n",
       "6707                 -2.111925               -2.088475              -0.004385   \n",
       "170355               -2.111925               -2.088475              -0.004385   \n",
       "246568               -2.111925               -2.088475              -0.004385   \n",
       "49718                -2.111925               -2.088475              -0.004385   \n",
       "\n",
       "        roll_mean_price_29_10  roll_mean_price_27_8  roll_mean_price_25_6  \\\n",
       "sku                                                                         \n",
       "464801               0.000191             -0.004176              -0.00350   \n",
       "645793               0.000191             -0.004176              -0.00350   \n",
       "99516               -0.004378             -0.004405              -0.00362   \n",
       "538100              -0.004378             -0.004405              -0.00362   \n",
       "557191              -0.004378             -0.004405              -0.00362   \n",
       "...                       ...                   ...                   ...   \n",
       "129187              -0.004378             -0.004405              -0.00362   \n",
       "6707                -0.004378             -0.004405              -0.00362   \n",
       "170355              -0.004378             -0.004405              -0.00362   \n",
       "246568              -0.004378             -0.004405              -0.00362   \n",
       "49718               -0.004378             -0.004405              -0.00362   \n",
       "\n",
       "        roll_mean_price_24_5  roll_sum_zero_sold_30_11  \\\n",
       "sku                                                      \n",
       "464801             -0.003556                  0.527432   \n",
       "645793             -0.003556                 -0.073491   \n",
       "99516              -0.003677                 -0.474106   \n",
       "538100             -0.003677                 -0.874720   \n",
       "557191             -0.003677                  0.928047   \n",
       "...                      ...                       ...   \n",
       "129187             -0.003677                 -2.677488   \n",
       "6707               -0.003677                 -2.677488   \n",
       "170355             -0.003677                 -2.677488   \n",
       "246568             -0.003677                 -2.677488   \n",
       "49718              -0.003677                 -2.677488   \n",
       "\n",
       "        roll_sum_zero_sold_24_5  \\\n",
       "sku                               \n",
       "464801                 0.314442   \n",
       "645793                -0.266020   \n",
       "99516                 -0.652995   \n",
       "538100                -0.459507   \n",
       "557191                 0.894904   \n",
       "...                         ...   \n",
       "129187                -2.781355   \n",
       "6707                  -2.781355   \n",
       "170355                -2.781355   \n",
       "246568                -2.781355   \n",
       "49718                 -2.781355   \n",
       "\n",
       "        roll_sum_shipping_logistic_type_fulfillment_30_11  \\\n",
       "sku                                                         \n",
       "464801                                           0.479567   \n",
       "645793                                           0.479567   \n",
       "99516                                           -0.792469   \n",
       "538100                                           0.479567   \n",
       "557191                                           0.479567   \n",
       "...                                                   ...   \n",
       "129187                                          -2.700524   \n",
       "6707                                            -2.700524   \n",
       "170355                                          -2.700524   \n",
       "246568                                          -2.541519   \n",
       "49718                                           -2.700524   \n",
       "\n",
       "        roll_sum_shipping_logistic_type_cross_docking_30_11  \\\n",
       "sku                                                           \n",
       "464801                                          -0.297576     \n",
       "645793                                          -0.297576     \n",
       "99516                                           -0.297576     \n",
       "538100                                          -0.297576     \n",
       "557191                                          -0.297576     \n",
       "...                                                   ...     \n",
       "129187                                          -0.297576     \n",
       "6707                                            -0.060230     \n",
       "170355                                          -0.297576     \n",
       "246568                                          -0.297576     \n",
       "49718                                           -0.297576     \n",
       "\n",
       "        roll_sum_shipping_logistic_type_fulfillment_28_9  \\\n",
       "sku                                                        \n",
       "464801                                          0.494024   \n",
       "645793                                          0.494024   \n",
       "99516                                          -0.744423   \n",
       "538100                                          0.494024   \n",
       "557191                                          0.494024   \n",
       "...                                                  ...   \n",
       "129187                                         -2.602093   \n",
       "6707                                           -2.602093   \n",
       "170355                                         -2.602093   \n",
       "246568                                         -2.602093   \n",
       "49718                                          -2.602093   \n",
       "\n",
       "        roll_sum_shipping_logistic_type_fulfillment_26_7  \\\n",
       "sku                                                        \n",
       "464801                                           0.50752   \n",
       "645793                                           0.50752   \n",
       "99516                                           -0.69828   \n",
       "538100                                           0.50752   \n",
       "557191                                           0.50752   \n",
       "...                                                  ...   \n",
       "129187                                          -2.50698   \n",
       "6707                                            -2.50698   \n",
       "170355                                          -2.50698   \n",
       "246568                                          -2.50698   \n",
       "49718                                           -2.50698   \n",
       "\n",
       "        roll_sum_shipping_logistic_type_fulfillment_25_6  \\\n",
       "sku                                                        \n",
       "464801                                          0.514066   \n",
       "645793                                          0.514066   \n",
       "99516                                          -0.676118   \n",
       "538100                                          0.514066   \n",
       "557191                                          0.514066   \n",
       "...                                                  ...   \n",
       "129187                                         -2.461394   \n",
       "6707                                           -2.461394   \n",
       "170355                                         -2.461394   \n",
       "246568                                         -2.461394   \n",
       "49718                                          -2.461394   \n",
       "\n",
       "        roll_sum_shipping_logistic_type_fulfillment_24_5  \\\n",
       "sku                                                        \n",
       "464801                                          0.520863   \n",
       "645793                                          0.520863   \n",
       "99516                                          -0.654889   \n",
       "538100                                          0.520863   \n",
       "557191                                          0.520863   \n",
       "...                                                  ...   \n",
       "129187                                         -2.418515   \n",
       "6707                                           -2.418515   \n",
       "170355                                         -2.418515   \n",
       "246568                                         -2.418515   \n",
       "49718                                          -2.418515   \n",
       "\n",
       "        roll_sum_shipping_payment_free_shipping_28_9  target_stock  \n",
       "sku                                                                 \n",
       "464801                                      1.490386     -0.124437  \n",
       "645793                                      1.490386     -0.116708  \n",
       "99516                                      -0.686343     -0.085792  \n",
       "538100                                     -0.686343     -0.085792  \n",
       "557191                                     -0.686343     -0.070334  \n",
       "...                                              ...           ...  \n",
       "129187                                     -0.686343     -0.085792  \n",
       "6707                                       -0.686343      0.084246  \n",
       "170355                                     -0.686343     -0.124437  \n",
       "246568                                     -0.686343     -0.132166  \n",
       "49718                                      -0.686343      0.006956  \n",
       "\n",
       "[551472 rows x 40 columns]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test = pd.read_feather('X_test_standard.f').set_index('sku')\n",
    "X_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "4488da75",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "sold_quantity_17                                       0\n",
       "sold_quantity_18                                       0\n",
       "sold_quantity_19                                       0\n",
       "sold_quantity_20                                       0\n",
       "sold_quantity_21                                       0\n",
       "sold_quantity_25                                       0\n",
       "sold_quantity_26                                       0\n",
       "sold_quantity_27                                       0\n",
       "sold_quantity_28                                       0\n",
       "sold_quantity_29                                       0\n",
       "sold_quantity_30                                       0\n",
       "minutes_active_29                                      0\n",
       "minutes_active_30                                      0\n",
       "roll_mean_sold_30_11                                   0\n",
       "roll_mean_sold_29_10                                   0\n",
       "roll_mean_sold_28_9                                    0\n",
       "roll_mean_sold_27_8                                    0\n",
       "roll_mean_sold_26_7                                    0\n",
       "roll_mean_sold_25_6                                    0\n",
       "roll_mean_sold_24_5                                    0\n",
       "roll_mean_minutes_30_11                                0\n",
       "roll_mean_minutes_29_10                                0\n",
       "roll_mean_minutes_26_7                                 0\n",
       "roll_mean_minutes_25_6                                 0\n",
       "roll_mean_minutes_24_5                                 0\n",
       "roll_mean_price_30_11                                  0\n",
       "roll_mean_price_29_10                                  0\n",
       "roll_mean_price_27_8                                   0\n",
       "roll_mean_price_25_6                                   0\n",
       "roll_mean_price_24_5                                   0\n",
       "roll_sum_zero_sold_30_11                               0\n",
       "roll_sum_zero_sold_24_5                                0\n",
       "roll_sum_shipping_logistic_type_fulfillment_30_11      0\n",
       "roll_sum_shipping_logistic_type_cross_docking_30_11    0\n",
       "roll_sum_shipping_logistic_type_fulfillment_28_9       0\n",
       "roll_sum_shipping_logistic_type_fulfillment_26_7       0\n",
       "roll_sum_shipping_logistic_type_fulfillment_25_6       0\n",
       "roll_sum_shipping_logistic_type_fulfillment_24_5       0\n",
       "roll_sum_shipping_payment_free_shipping_28_9           0\n",
       "target_stock                                           0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# how many NAs \n",
    "X_test.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "a40aa45e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 8.91 s, sys: 1.66 s, total: 10.6 s\n",
      "Wall time: 7.68 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "final_y_pred_nn = model.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "1c31b629",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[1.36590451e-02, 2.57467665e-02, 3.54938060e-02, ...,\n",
       "        1.52956424e-02, 1.17993020e-02, 8.01451690e-03],\n",
       "       [4.93532559e-03, 1.69685632e-02, 2.89007910e-02, ...,\n",
       "        1.37665495e-02, 8.68697651e-03, 5.17488038e-03],\n",
       "       [2.51003087e-01, 2.27738529e-01, 1.55520961e-01, ...,\n",
       "        7.62241601e-04, 6.88130851e-04, 6.50360773e-04],\n",
       "       ...,\n",
       "       [3.40627943e-04, 9.49128706e-04, 3.37733724e-03, ...,\n",
       "        6.90114796e-02, 4.41541225e-02, 4.94495146e-02],\n",
       "       [1.14059187e-02, 1.20163914e-02, 2.15291791e-02, ...,\n",
       "        4.70974743e-02, 3.17597352e-02, 3.80777493e-02],\n",
       "       [7.57517782e-09, 5.12989391e-06, 9.77990785e-05, ...,\n",
       "        6.69068173e-02, 4.17152122e-02, 2.97623500e-02]], dtype=float32)"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "final_y_pred_nn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "2ce53334",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(551472, 30)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "final_y_pred_nn.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bb8ffc9d",
   "metadata": {},
   "source": [
    "## Creating DataFrame with predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "759734a9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 160 ms, sys: 45.3 ms, total: 205 ms\n",
      "Wall time: 211 ms\n"
     ]
    },
    {
     "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>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>9</th>\n",
       "      <th>10</th>\n",
       "      <th>11</th>\n",
       "      <th>12</th>\n",
       "      <th>13</th>\n",
       "      <th>14</th>\n",
       "      <th>15</th>\n",
       "      <th>16</th>\n",
       "      <th>17</th>\n",
       "      <th>18</th>\n",
       "      <th>19</th>\n",
       "      <th>20</th>\n",
       "      <th>21</th>\n",
       "      <th>22</th>\n",
       "      <th>23</th>\n",
       "      <th>24</th>\n",
       "      <th>25</th>\n",
       "      <th>26</th>\n",
       "      <th>27</th>\n",
       "      <th>28</th>\n",
       "      <th>29</th>\n",
       "      <th>30</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1.365905e-02</td>\n",
       "      <td>0.025747</td>\n",
       "      <td>0.035494</td>\n",
       "      <td>0.039322</td>\n",
       "      <td>0.038123</td>\n",
       "      <td>0.040007</td>\n",
       "      <td>0.049261</td>\n",
       "      <td>0.056385</td>\n",
       "      <td>0.056271</td>\n",
       "      <td>0.057955</td>\n",
       "      <td>0.045832</td>\n",
       "      <td>0.036673</td>\n",
       "      <td>0.038136</td>\n",
       "      <td>0.044078</td>\n",
       "      <td>0.044692</td>\n",
       "      <td>0.044576</td>\n",
       "      <td>0.041030</td>\n",
       "      <td>0.036218</td>\n",
       "      <td>0.028938</td>\n",
       "      <td>0.026893</td>\n",
       "      <td>0.028980</td>\n",
       "      <td>0.031269</td>\n",
       "      <td>0.027104</td>\n",
       "      <td>0.026480</td>\n",
       "      <td>0.019834</td>\n",
       "      <td>0.015685</td>\n",
       "      <td>0.016250</td>\n",
       "      <td>0.015296</td>\n",
       "      <td>0.011799</td>\n",
       "      <td>0.008015</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>4.935326e-03</td>\n",
       "      <td>0.016969</td>\n",
       "      <td>0.028901</td>\n",
       "      <td>0.037807</td>\n",
       "      <td>0.039920</td>\n",
       "      <td>0.041487</td>\n",
       "      <td>0.053913</td>\n",
       "      <td>0.063660</td>\n",
       "      <td>0.059959</td>\n",
       "      <td>0.065777</td>\n",
       "      <td>0.049301</td>\n",
       "      <td>0.039782</td>\n",
       "      <td>0.043458</td>\n",
       "      <td>0.046389</td>\n",
       "      <td>0.045606</td>\n",
       "      <td>0.046141</td>\n",
       "      <td>0.043256</td>\n",
       "      <td>0.035481</td>\n",
       "      <td>0.029220</td>\n",
       "      <td>0.027716</td>\n",
       "      <td>0.027080</td>\n",
       "      <td>0.029754</td>\n",
       "      <td>0.026941</td>\n",
       "      <td>0.023869</td>\n",
       "      <td>0.017534</td>\n",
       "      <td>0.013567</td>\n",
       "      <td>0.013947</td>\n",
       "      <td>0.013767</td>\n",
       "      <td>0.008687</td>\n",
       "      <td>0.005175</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2.510031e-01</td>\n",
       "      <td>0.227739</td>\n",
       "      <td>0.155521</td>\n",
       "      <td>0.101356</td>\n",
       "      <td>0.066518</td>\n",
       "      <td>0.045883</td>\n",
       "      <td>0.035650</td>\n",
       "      <td>0.027778</td>\n",
       "      <td>0.017459</td>\n",
       "      <td>0.014165</td>\n",
       "      <td>0.009412</td>\n",
       "      <td>0.006965</td>\n",
       "      <td>0.005200</td>\n",
       "      <td>0.005197</td>\n",
       "      <td>0.004850</td>\n",
       "      <td>0.004495</td>\n",
       "      <td>0.003310</td>\n",
       "      <td>0.002592</td>\n",
       "      <td>0.001926</td>\n",
       "      <td>0.001764</td>\n",
       "      <td>0.001735</td>\n",
       "      <td>0.001779</td>\n",
       "      <td>0.001513</td>\n",
       "      <td>0.001317</td>\n",
       "      <td>0.001136</td>\n",
       "      <td>0.000750</td>\n",
       "      <td>0.000887</td>\n",
       "      <td>0.000762</td>\n",
       "      <td>0.000688</td>\n",
       "      <td>0.000650</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>8.849484e-03</td>\n",
       "      <td>0.046022</td>\n",
       "      <td>0.076318</td>\n",
       "      <td>0.075585</td>\n",
       "      <td>0.091568</td>\n",
       "      <td>0.067799</td>\n",
       "      <td>0.070735</td>\n",
       "      <td>0.076430</td>\n",
       "      <td>0.058680</td>\n",
       "      <td>0.055814</td>\n",
       "      <td>0.046874</td>\n",
       "      <td>0.038594</td>\n",
       "      <td>0.032738</td>\n",
       "      <td>0.032110</td>\n",
       "      <td>0.029493</td>\n",
       "      <td>0.028290</td>\n",
       "      <td>0.024404</td>\n",
       "      <td>0.019061</td>\n",
       "      <td>0.015200</td>\n",
       "      <td>0.014644</td>\n",
       "      <td>0.014106</td>\n",
       "      <td>0.014195</td>\n",
       "      <td>0.012362</td>\n",
       "      <td>0.011267</td>\n",
       "      <td>0.008715</td>\n",
       "      <td>0.006940</td>\n",
       "      <td>0.007207</td>\n",
       "      <td>0.006560</td>\n",
       "      <td>0.005241</td>\n",
       "      <td>0.004200</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2.354006e-03</td>\n",
       "      <td>0.009160</td>\n",
       "      <td>0.017196</td>\n",
       "      <td>0.021125</td>\n",
       "      <td>0.027195</td>\n",
       "      <td>0.028443</td>\n",
       "      <td>0.036530</td>\n",
       "      <td>0.044038</td>\n",
       "      <td>0.047703</td>\n",
       "      <td>0.050473</td>\n",
       "      <td>0.044839</td>\n",
       "      <td>0.039908</td>\n",
       "      <td>0.039343</td>\n",
       "      <td>0.046460</td>\n",
       "      <td>0.051257</td>\n",
       "      <td>0.050585</td>\n",
       "      <td>0.048646</td>\n",
       "      <td>0.042794</td>\n",
       "      <td>0.035981</td>\n",
       "      <td>0.033048</td>\n",
       "      <td>0.040338</td>\n",
       "      <td>0.038613</td>\n",
       "      <td>0.037519</td>\n",
       "      <td>0.038116</td>\n",
       "      <td>0.029414</td>\n",
       "      <td>0.024535</td>\n",
       "      <td>0.023605</td>\n",
       "      <td>0.022392</td>\n",
       "      <td>0.016929</td>\n",
       "      <td>0.011461</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>551467</th>\n",
       "      <td>7.769506e-06</td>\n",
       "      <td>0.000133</td>\n",
       "      <td>0.000647</td>\n",
       "      <td>0.001741</td>\n",
       "      <td>0.002555</td>\n",
       "      <td>0.003839</td>\n",
       "      <td>0.006263</td>\n",
       "      <td>0.008883</td>\n",
       "      <td>0.014120</td>\n",
       "      <td>0.016747</td>\n",
       "      <td>0.017827</td>\n",
       "      <td>0.014006</td>\n",
       "      <td>0.024499</td>\n",
       "      <td>0.021765</td>\n",
       "      <td>0.031059</td>\n",
       "      <td>0.034810</td>\n",
       "      <td>0.054788</td>\n",
       "      <td>0.060147</td>\n",
       "      <td>0.036715</td>\n",
       "      <td>0.039249</td>\n",
       "      <td>0.040851</td>\n",
       "      <td>0.112356</td>\n",
       "      <td>0.060007</td>\n",
       "      <td>0.063671</td>\n",
       "      <td>0.059593</td>\n",
       "      <td>0.046720</td>\n",
       "      <td>0.071724</td>\n",
       "      <td>0.071208</td>\n",
       "      <td>0.045513</td>\n",
       "      <td>0.038556</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>551468</th>\n",
       "      <td>1.277076e-09</td>\n",
       "      <td>0.000003</td>\n",
       "      <td>0.000075</td>\n",
       "      <td>0.000597</td>\n",
       "      <td>0.001508</td>\n",
       "      <td>0.003740</td>\n",
       "      <td>0.007394</td>\n",
       "      <td>0.011041</td>\n",
       "      <td>0.016934</td>\n",
       "      <td>0.021555</td>\n",
       "      <td>0.020089</td>\n",
       "      <td>0.018696</td>\n",
       "      <td>0.031858</td>\n",
       "      <td>0.034758</td>\n",
       "      <td>0.041881</td>\n",
       "      <td>0.052781</td>\n",
       "      <td>0.063458</td>\n",
       "      <td>0.067605</td>\n",
       "      <td>0.039217</td>\n",
       "      <td>0.050993</td>\n",
       "      <td>0.053279</td>\n",
       "      <td>0.106113</td>\n",
       "      <td>0.060825</td>\n",
       "      <td>0.057431</td>\n",
       "      <td>0.054456</td>\n",
       "      <td>0.042363</td>\n",
       "      <td>0.052732</td>\n",
       "      <td>0.050272</td>\n",
       "      <td>0.024821</td>\n",
       "      <td>0.013526</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>551469</th>\n",
       "      <td>3.406279e-04</td>\n",
       "      <td>0.000949</td>\n",
       "      <td>0.003377</td>\n",
       "      <td>0.005541</td>\n",
       "      <td>0.007497</td>\n",
       "      <td>0.007422</td>\n",
       "      <td>0.011217</td>\n",
       "      <td>0.015393</td>\n",
       "      <td>0.026411</td>\n",
       "      <td>0.025877</td>\n",
       "      <td>0.024767</td>\n",
       "      <td>0.018509</td>\n",
       "      <td>0.024923</td>\n",
       "      <td>0.027040</td>\n",
       "      <td>0.036168</td>\n",
       "      <td>0.035961</td>\n",
       "      <td>0.059571</td>\n",
       "      <td>0.054977</td>\n",
       "      <td>0.030946</td>\n",
       "      <td>0.035764</td>\n",
       "      <td>0.035587</td>\n",
       "      <td>0.106989</td>\n",
       "      <td>0.045978</td>\n",
       "      <td>0.052622</td>\n",
       "      <td>0.047822</td>\n",
       "      <td>0.027593</td>\n",
       "      <td>0.068143</td>\n",
       "      <td>0.069011</td>\n",
       "      <td>0.044154</td>\n",
       "      <td>0.049450</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>551470</th>\n",
       "      <td>1.140592e-02</td>\n",
       "      <td>0.012016</td>\n",
       "      <td>0.021529</td>\n",
       "      <td>0.024266</td>\n",
       "      <td>0.022996</td>\n",
       "      <td>0.019841</td>\n",
       "      <td>0.026137</td>\n",
       "      <td>0.029340</td>\n",
       "      <td>0.046717</td>\n",
       "      <td>0.040069</td>\n",
       "      <td>0.033648</td>\n",
       "      <td>0.022142</td>\n",
       "      <td>0.030140</td>\n",
       "      <td>0.028988</td>\n",
       "      <td>0.035878</td>\n",
       "      <td>0.033980</td>\n",
       "      <td>0.053581</td>\n",
       "      <td>0.048542</td>\n",
       "      <td>0.027439</td>\n",
       "      <td>0.028826</td>\n",
       "      <td>0.027634</td>\n",
       "      <td>0.081862</td>\n",
       "      <td>0.034885</td>\n",
       "      <td>0.037301</td>\n",
       "      <td>0.034852</td>\n",
       "      <td>0.019932</td>\n",
       "      <td>0.049117</td>\n",
       "      <td>0.047097</td>\n",
       "      <td>0.031760</td>\n",
       "      <td>0.038078</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>551471</th>\n",
       "      <td>7.575178e-09</td>\n",
       "      <td>0.000005</td>\n",
       "      <td>0.000098</td>\n",
       "      <td>0.000575</td>\n",
       "      <td>0.001372</td>\n",
       "      <td>0.002857</td>\n",
       "      <td>0.004825</td>\n",
       "      <td>0.007734</td>\n",
       "      <td>0.010714</td>\n",
       "      <td>0.014117</td>\n",
       "      <td>0.015152</td>\n",
       "      <td>0.013367</td>\n",
       "      <td>0.022291</td>\n",
       "      <td>0.024964</td>\n",
       "      <td>0.033088</td>\n",
       "      <td>0.043563</td>\n",
       "      <td>0.052722</td>\n",
       "      <td>0.057843</td>\n",
       "      <td>0.038984</td>\n",
       "      <td>0.044191</td>\n",
       "      <td>0.048500</td>\n",
       "      <td>0.117592</td>\n",
       "      <td>0.063294</td>\n",
       "      <td>0.066756</td>\n",
       "      <td>0.058747</td>\n",
       "      <td>0.050030</td>\n",
       "      <td>0.068235</td>\n",
       "      <td>0.066907</td>\n",
       "      <td>0.041715</td>\n",
       "      <td>0.029762</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>551472 rows × 30 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                  1         2         3         4         5         6   \\\n",
       "0       1.365905e-02  0.025747  0.035494  0.039322  0.038123  0.040007   \n",
       "1       4.935326e-03  0.016969  0.028901  0.037807  0.039920  0.041487   \n",
       "2       2.510031e-01  0.227739  0.155521  0.101356  0.066518  0.045883   \n",
       "3       8.849484e-03  0.046022  0.076318  0.075585  0.091568  0.067799   \n",
       "4       2.354006e-03  0.009160  0.017196  0.021125  0.027195  0.028443   \n",
       "...              ...       ...       ...       ...       ...       ...   \n",
       "551467  7.769506e-06  0.000133  0.000647  0.001741  0.002555  0.003839   \n",
       "551468  1.277076e-09  0.000003  0.000075  0.000597  0.001508  0.003740   \n",
       "551469  3.406279e-04  0.000949  0.003377  0.005541  0.007497  0.007422   \n",
       "551470  1.140592e-02  0.012016  0.021529  0.024266  0.022996  0.019841   \n",
       "551471  7.575178e-09  0.000005  0.000098  0.000575  0.001372  0.002857   \n",
       "\n",
       "              7         8         9         10        11        12        13  \\\n",
       "0       0.049261  0.056385  0.056271  0.057955  0.045832  0.036673  0.038136   \n",
       "1       0.053913  0.063660  0.059959  0.065777  0.049301  0.039782  0.043458   \n",
       "2       0.035650  0.027778  0.017459  0.014165  0.009412  0.006965  0.005200   \n",
       "3       0.070735  0.076430  0.058680  0.055814  0.046874  0.038594  0.032738   \n",
       "4       0.036530  0.044038  0.047703  0.050473  0.044839  0.039908  0.039343   \n",
       "...          ...       ...       ...       ...       ...       ...       ...   \n",
       "551467  0.006263  0.008883  0.014120  0.016747  0.017827  0.014006  0.024499   \n",
       "551468  0.007394  0.011041  0.016934  0.021555  0.020089  0.018696  0.031858   \n",
       "551469  0.011217  0.015393  0.026411  0.025877  0.024767  0.018509  0.024923   \n",
       "551470  0.026137  0.029340  0.046717  0.040069  0.033648  0.022142  0.030140   \n",
       "551471  0.004825  0.007734  0.010714  0.014117  0.015152  0.013367  0.022291   \n",
       "\n",
       "              14        15        16        17        18        19        20  \\\n",
       "0       0.044078  0.044692  0.044576  0.041030  0.036218  0.028938  0.026893   \n",
       "1       0.046389  0.045606  0.046141  0.043256  0.035481  0.029220  0.027716   \n",
       "2       0.005197  0.004850  0.004495  0.003310  0.002592  0.001926  0.001764   \n",
       "3       0.032110  0.029493  0.028290  0.024404  0.019061  0.015200  0.014644   \n",
       "4       0.046460  0.051257  0.050585  0.048646  0.042794  0.035981  0.033048   \n",
       "...          ...       ...       ...       ...       ...       ...       ...   \n",
       "551467  0.021765  0.031059  0.034810  0.054788  0.060147  0.036715  0.039249   \n",
       "551468  0.034758  0.041881  0.052781  0.063458  0.067605  0.039217  0.050993   \n",
       "551469  0.027040  0.036168  0.035961  0.059571  0.054977  0.030946  0.035764   \n",
       "551470  0.028988  0.035878  0.033980  0.053581  0.048542  0.027439  0.028826   \n",
       "551471  0.024964  0.033088  0.043563  0.052722  0.057843  0.038984  0.044191   \n",
       "\n",
       "              21        22        23        24        25        26        27  \\\n",
       "0       0.028980  0.031269  0.027104  0.026480  0.019834  0.015685  0.016250   \n",
       "1       0.027080  0.029754  0.026941  0.023869  0.017534  0.013567  0.013947   \n",
       "2       0.001735  0.001779  0.001513  0.001317  0.001136  0.000750  0.000887   \n",
       "3       0.014106  0.014195  0.012362  0.011267  0.008715  0.006940  0.007207   \n",
       "4       0.040338  0.038613  0.037519  0.038116  0.029414  0.024535  0.023605   \n",
       "...          ...       ...       ...       ...       ...       ...       ...   \n",
       "551467  0.040851  0.112356  0.060007  0.063671  0.059593  0.046720  0.071724   \n",
       "551468  0.053279  0.106113  0.060825  0.057431  0.054456  0.042363  0.052732   \n",
       "551469  0.035587  0.106989  0.045978  0.052622  0.047822  0.027593  0.068143   \n",
       "551470  0.027634  0.081862  0.034885  0.037301  0.034852  0.019932  0.049117   \n",
       "551471  0.048500  0.117592  0.063294  0.066756  0.058747  0.050030  0.068235   \n",
       "\n",
       "              28        29        30  \n",
       "0       0.015296  0.011799  0.008015  \n",
       "1       0.013767  0.008687  0.005175  \n",
       "2       0.000762  0.000688  0.000650  \n",
       "3       0.006560  0.005241  0.004200  \n",
       "4       0.022392  0.016929  0.011461  \n",
       "...          ...       ...       ...  \n",
       "551467  0.071208  0.045513  0.038556  \n",
       "551468  0.050272  0.024821  0.013526  \n",
       "551469  0.069011  0.044154  0.049450  \n",
       "551470  0.047097  0.031760  0.038078  \n",
       "551471  0.066907  0.041715  0.029762  \n",
       "\n",
       "[551472 rows x 30 columns]"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "X_test_sub = pd.DataFrame(final_y_pred_nn, columns=range(1, 31))\n",
    "X_test_sub = X_test_sub.astype(float)\n",
    "X_test_sub"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "44ee68d1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# check probabilites for random skus\n",
    "for _ in range(1, 5):\n",
    "    random_sku = X_test_sub.sample(1)\n",
    "    plt.plot(range(1, 31), random_sku.values.squeeze())\n",
    "plt.grid();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "3b40b0a9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.0    1\n",
       "1.0    1\n",
       "1.0    1\n",
       "1.0    1\n",
       "1.0    1\n",
       "      ..\n",
       "1.0    1\n",
       "1.0    1\n",
       "1.0    1\n",
       "1.0    1\n",
       "1.0    1\n",
       "Length: 85498, dtype: int64"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test_sub.sum(axis=1).value_counts().sort_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "3acce227",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA2oAAAHdCAYAAABse4rvAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjQuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/MnkTPAAAACXBIWXMAAAsTAAALEwEAmpwYAAAbQ0lEQVR4nO3df4zk913f8dc7vgTaEALi2iuyTc4qTosJpKFbJxTULE1anVPJVsUP2SUNpianSiSiBaEagRIU/mmIIBWSk3CAZYKEjUkRPdXmjEQydQVxZEcJbmzL6cmh8Rkkg/OjWkchmLz7x26qZXN3O5eb785nbh4P6aSdma9n3pp56+ynZ+a71d0BAABgHM9b9gAAAAD8TUINAABgMEINAABgMEINAABgMEINAABgMEINAABgMEsNtaq6vaqerqqPzXHsO6vqozt/Pl5VnzmAEQEAAA5cLfP3qFXVP0uyleS93f2yC/jn3pzkFd397yYbDgAAYEmW+o5ad9+f5FO7r6uqv19Vp6rqw1X1P6vqH57lH70pyZ0HMiQAAMABO7TsAc7iRJJ/393/u6pemeRdSf75l26sqpckuSrJ+5c0HwAAwKSGCrWq+pok/zTJb1fVl67+qj2H3Zjkfd391wc5GwAAwEEZKtSy/VHMz3T3PzrPMTcm+dGDGQcAAODgDXV6/u7+v0k+UVXfnyS17eVfun3n+2pfn+SDSxoRAABgcss+Pf+d2Y6uf1BVZ6rqliQ/mOSWqvrjJI8kuWHXP3Jjkrt6maeqBAAAmNhST88PAADAlxvqo48AAAAINQAAgOEs7ayPhw8f7qNHjy7r4c/p2WefzQtf+MJlj8Elyn4xNTvGlOwXU7JfTGnU/frwhz/8F939d85229JC7ejRo3nooYeW9fDnNJvNsrm5uewxuETZL6Zmx5iS/WJK9ospjbpfVfV/znWbjz4CAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAM5tCyBxjNzaeeTU7ds+wxVsaf/Od/tewRAADgkrPvO2pVdXtVPV1VHzvPMZtV9dGqeqSq/sdiRwQAAFgv83z08Y4kx851Y1V9XZJ3Jbm+u781yfcvZDIAAIA1tW+odff9ST51nkP+TZLf6e5P7hz/9IJmAwAAWEuLOJnIS5N8fVXNqurDVfWGBdwnAADA2lrEyUQOJfnHSV6T5G8l+WBVPdDdH997YFUdT3I8SY4cOZLZbLaAh2eZvIYXZmtry3PGpOwYU7JfTMl+MaVV3K9FhNqZJM9097NJnq2q+5O8PMmXhVp3n0hyIkk2NjZ6c3NzAQ+/YM74eEGGfA0HNpvNPGdMyo4xJfvFlOwXU1rF/VrERx//W5LvrqpDVfW3k7wyyWMLuF8AAIC1tO87alV1Z5LNJIer6kyStyZ5fpJ093u6+7GqOpXk4SRfTPKr3X3OU/kDAABwfvuGWnffNMcx70jyjoVMBAAAsOYW8dFHAAAAFkioAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADEaoAQAADGbfUKuq26vq6ar62D7H/ZOqeq6qvm9x4wEAAKyfed5RuyPJsfMdUFWXJXl7kt9fwEwAAABrbd9Q6+77k3xqn8PenOS/Jnl6EUMBAACss4v+jlpVXZ7kXyd598WPAwAAwKEF3Md/SfKfuvuLVXXeA6vqeJLjSXLkyJHMZrMFPDzL5DW8MFtbW54zJmXHmJL9Ykr2iymt4n4tItQ2kty1E2mHk7yuqp7r7t/de2B3n0hyIkk2NjZ6c3NzAQ+/YKfuWfYEK2XI13Bgs9nMc8ak7BhTsl9MyX4xpVXcr4sOte6+6ks/V9UdSf772SINAACA+ewbalV1Z5LNJIer6kyStyZ5fpJ093smnQ4AAGAN7Rtq3X3TvHfW3Tdf1DQAAABc/FkfAQAAWCyhBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMBihBgAAMJh9Q62qbq+qp6vqY+e4/Qer6uGq+l9V9UdV9fLFjwkAALA+5nlH7Y4kx85z+yeSvLq7vy3JzyU5sYC5AAAA1tah/Q7o7vur6uh5bv+jXRcfSHLFAuYCAABYW/uG2gW6JcnvnevGqjqe5HiSHDlyJLPZbMEPz0HzGl6Yra0tzxmTsmNMyX4xJfvFlFZxvxYWalX1PdkOte8+1zHdfSI7H43c2Njozc3NRT384py6Z9kTrJQhX8OBzWYzzxmTsmNMyX4xJfvFlFZxvxYSalX17Ul+Ncl13f3MIu4TAABgXV306fmr6puS/E6Sf9vdH7/4kQAAANbbvu+oVdWdSTaTHK6qM0nemuT5SdLd70nyliTfkORdVZUkz3X3xlQDAwAAXOrmOevjTfvc/iNJfmRhEwEAAKy5i/7oIwAAAIsl1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAazb6hV1e1V9XRVfewct1dV/VJVna6qh6vqOxY/JgAAwPqY5x21O5IcO8/t1yW5eufP8STvvvixAAAA1te+odbd9yf51HkOuSHJe3vbA0m+rqq+cVEDAgAArJtDC7iPy5M8uevymZ3r/mzvgVV1PNvvuuXIkSOZzWYLeHiWyWt4Yba2tjxnTMqOMSX7xZTsF1Naxf1aRKjNrbtPJDmRJBsbG725uXmQDz+fU/cse4KVMuRrOLDZbOY5Y1J2jCnZL6Zkv5jSKu7XIs76+FSSK3ddvmLnOgAAAL4Ciwi1k0nesHP2x1cl+Wx3f9nHHgEAAJjPvh99rKo7k2wmOVxVZ5K8Ncnzk6S735Pk3iSvS3I6yeeS/PBUwwIAAKyDfUOtu2/a5/ZO8qMLmwgAAGDNLeKjjwAAACyQUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABiMUAMAABjMXKFWVceq6vGqOl1Vt57l9m+qqg9U1Ueq6uGqet3iRwUAAFgP+4ZaVV2W5LYk1yW5JslNVXXNnsN+Jsnd3f2KJDcmedeiBwUAAFgX87yjdm2S0939RHd/IcldSW7Yc0wn+dqdn1+c5E8XNyIAAMB6OTTHMZcneXLX5TNJXrnnmJ9N8vtV9eYkL0zy2rPdUVUdT3I8SY4cOZLZbHaB4zIar+GF2dra8pwxKTvGlOwXU7JfTGkV92ueUJvHTUnu6O5fqKrvTPIbVfWy7v7i7oO6+0SSE0mysbHRm5ubC3r4BTp1z7InWClDvoYDm81mnjMmZceYkv1iSvaLKa3ifs3z0cenkly56/IVO9ftdkuSu5Okuz+Y5KuTHF7EgAAAAOtmnlB7MMnVVXVVVb0g2ycLObnnmE8meU2SVNW3ZDvU/nyRgwIAAKyLfUOtu59L8qYk9yV5LNtnd3ykqt5WVdfvHPYTSd5YVX+c5M4kN3d3TzU0AADApWyu76h1971J7t1z3Vt2/fxoku9a7GgAAADraa5feA0AAMDBEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDEWoAAACDmSvUqupYVT1eVaer6tZzHPMDVfVoVT1SVb+52DEBAADWx6H9Dqiqy5LcluRfJDmT5MGqOtndj+465uokP5Xku7r701X1d6caGAAA4FI3zztq1yY53d1PdPcXktyV5IY9x7wxyW3d/ekk6e6nFzsmAADA+pgn1C5P8uSuy2d2rtvtpUleWlV/WFUPVNWxRQ0IAACwbvb96OMF3M/VSTaTXJHk/qr6tu7+zO6Dqup4kuNJcuTIkcxmswU9PMviNbwwW1tbnjMmZceYkv1iSvaLKa3ifs0Tak8luXLX5St2rtvtTJIPdfdfJflEVX082+H24O6DuvtEkhNJsrGx0Zubm1/h2BM6dc+yJ1gpQ76GA5vNZp4zJmXHmJL9Ykr2iymt4n7N89HHB5NcXVVXVdULktyY5OSeY3432++mpaoOZ/ujkE8sbkwAAID1sW+odfdzSd6U5L4kjyW5u7sfqaq3VdX1O4fdl+SZqno0yQeS/GR3PzPV0AAAAJeyub6j1t33Jrl3z3Vv2fVzJ/nxnT8AAABchLl+4TUAAAAHR6gBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMRqgBAAAMZq5Qq6pjVfV4VZ2uqlvPc9z3VlVX1cbiRgQAAFgv+4ZaVV2W5LYk1yW5JslNVXXNWY57UZIfS/KhRQ8JAACwTuZ5R+3aJKe7+4nu/kKSu5LccJbjfi7J25N8foHzAQAArJ15Qu3yJE/uunxm57r/r6q+I8mV3X3PAmcDAABYS4cu9g6q6nlJfjHJzXMcezzJ8SQ5cuRIZrPZxT48S+Y1vDBbW1ueMyZlx5iS/WJK9ospreJ+zRNqTyW5ctflK3au+5IXJXlZkllVJcnfS3Kyqq7v7od231F3n0hyIkk2NjZ6c3PzK598Kqe8KXghhnwNBzabzTxnTMqOMSX7xZTsF1Naxf2a56OPDya5uqquqqoXJLkxyckv3djdn+3uw919tLuPJnkgyZdFGgAAAPPZN9S6+7kkb0pyX5LHktzd3Y9U1duq6vqpBwQAAFg3c31HrbvvTXLvnuveco5jNy9+LAAAgPU11y+8BgAA4OAINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMEINQAAgMHMFWpVdayqHq+q01V161lu//GqerSqHq6qP6iqlyx+VAAAgPWwb6hV1WVJbktyXZJrktxUVdfsOewjSTa6+9uTvC/Jzy96UAAAgHUxzztq1yY53d1PdPcXktyV5IbdB3T3B7r7czsXH0hyxWLHBAAAWB/zhNrlSZ7cdfnMznXnckuS37uYoQAAANbZoUXeWVW9PslGklef4/bjSY4nyZEjRzKbzRb58CyB1/DCbG1tec6YlB1jSvaLKdkvprSK+zVPqD2V5Mpdl6/Yue5vqKrXJvnpJK/u7r882x1194kkJ5JkY2OjNzc3L3Te6Z26Z9kTrJQhX8OBzWYzzxmTsmNMyX4xJfvFlFZxv+b56OODSa6uqquq6gVJbkxycvcBVfWKJL+c5PrufnrxYwIAAKyPfUOtu59L8qYk9yV5LMnd3f1IVb2tqq7fOewdSb4myW9X1Uer6uQ57g4AAIB9zPUdte6+N8m9e657y66fX7vguQAAANbWXL/wGgAAgIMj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAYj1AAAAAZzaNkDsNqO3nrPskdYKXcce+GyRwAAYAV4Rw0AAGAwQg0AAGAwc4VaVR2rqser6nRV3XqW27+qqn5r5/YPVdXRhU8KAACwJvYNtaq6LMltSa5Lck2Sm6rqmj2H3ZLk0939zUnemeTtix4UAABgXczzjtq1SU539xPd/YUkdyW5Yc8xNyT59Z2f35fkNVVVixsTAABgfcxz1sfLkzy56/KZJK881zHd/VxVfTbJNyT5i0UMCZeKm089m5xypkym48yiAHBpONDT81fV8STHdy5uVdXjB/n4czocgcl07BeT+p632zEmZb+Ykv1iSqPu10vOdcM8ofZUkit3Xb5i57qzHXOmqg4leXGSZ/beUXefSHJijsdcmqp6qLs3lj0Hlyb7xdTsGFOyX0zJfjGlVdyveb6j9mCSq6vqqqp6QZIbk5zcc8zJJD+08/P3JXl/d/fixgQAAFgf+76jtvOdszcluS/JZUlu7+5HquptSR7q7pNJfi3Jb1TV6SSfynbMAQAA8BWY6ztq3X1vknv3XPeWXT9/Psn3L3a0pRn6o5msPPvF1OwYU7JfTMl+MaWV26/yCUUAAICxzPMdNQAAAA7Q2oZaVR2rqser6nRV3XqW27+qqn5r5/YPVdXRJYzJippjv368qh6tqoer6g+q6pynZoW99tuvXcd9b1V1Va3UWa5Yrnn2q6p+YOfvsEeq6jcPekZW2xz/jvymqvpAVX1k59+Tr1vGnKyeqrq9qp6uqo+d4/aqql/a2b2Hq+o7DnrGC7GWoVZVlyW5Lcl1Sa5JclNVXbPnsFuSfLq7vznJO5O8/WCnZFXNuV8fSbLR3d+e5H1Jfv5gp2RVzblfqaoXJfmxJB862AlZZfPsV1VdneSnknxXd39rkv9w0HOyuub8O+xnktzd3a/I9gnq3nWwU7LC7khy7Dy3X5fk6p0/x5O8+wBm+oqtZagluTbJ6e5+oru/kOSuJDfsOeaGJL++8/P7krymquoAZ2R17btf3f2B7v7czsUHsv37CWEe8/z9lSQ/l+3/wfT5gxyOlTfPfr0xyW3d/ekk6e6nD3hGVts8O9ZJvnbn5xcn+dMDnI8V1t33Z/sM9OdyQ5L39rYHknxdVX3jwUx34dY11C5P8uSuy2d2rjvrMd39XJLPJvmGA5mOVTfPfu12S5Lfm3QiLiX77tfORzmu7O57DnIwLgnz/P310iQvrao/rKoHqup8//ca9ppnx342yeur6ky2zzr+5oMZjTVwof+NtlRznZ4fmEZVvT7JRpJXL3sWLg1V9bwkv5jk5iWPwqXrULY/NrSZ7U8D3F9V39bdn1nmUFxSbkpyR3f/QlV9Z7Z/V+/LuvuLyx4MDtK6vqP2VJIrd12+Yue6sx5TVYey/db7MwcyHatunv1KVb02yU8nub67//KAZmP17bdfL0rysiSzqvqTJK9KctIJRZjTPH9/nUlysrv/qrs/keTj2Q43mMc8O3ZLkruTpLs/mOSrkxw+kOm41M3132ijWNdQezDJ1VV1VVW9INtfVD2555iTSX5o5+fvS/L+9kvnmM+++1VVr0jyy9mONN/v4EKcd7+6+7Pdfbi7j3b30Wx/B/L67n5oOeOyYub59+PvZvvdtFTV4Wx/FPKJA5yR1TbPjn0yyWuSpKq+Jduh9ucHOiWXqpNJ3rBz9sdXJflsd//Zsoc6l7X86GN3P1dVb0pyX5LLktze3Y9U1duSPNTdJ5P8Wrbfaj+d7S8l3ri8iVklc+7XO5J8TZLf3jlHzSe7+/qlDc3KmHO/4Csy537dl+RfVtWjSf46yU92t0+cMJc5d+wnkvxKVf3HbJ9Y5Gb/s5x5VNWd2f4fSYd3vuP41iTPT5Lufk+2v/P4uiSnk3wuyQ8vZ9L5lL0HAAAYy7p+9BEAAGBYQg0AAGAwQg0AAGAwQg0AAGAwQg0AAGAwQg0AAGAwQg0AAGAwQg0AAGAw/w83k0y8TQEzZQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 1080x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.hist(X_test_sub.values.reshape(-1))\n",
    "plt.grid();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "4e52ed16",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "X_test_sub.idxmax(axis=1).plot.hist(bins=30)\n",
    "plt.xticks(range(1, 31))\n",
    "plt.grid();"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7b6158d4",
   "metadata": {},
   "source": [
    "Pretty different from LGBM predictions.   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "480f3459",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>9</th>\n",
       "      <th>10</th>\n",
       "      <th>11</th>\n",
       "      <th>12</th>\n",
       "      <th>13</th>\n",
       "      <th>14</th>\n",
       "      <th>15</th>\n",
       "      <th>16</th>\n",
       "      <th>17</th>\n",
       "      <th>18</th>\n",
       "      <th>19</th>\n",
       "      <th>20</th>\n",
       "      <th>21</th>\n",
       "      <th>22</th>\n",
       "      <th>23</th>\n",
       "      <th>24</th>\n",
       "      <th>25</th>\n",
       "      <th>26</th>\n",
       "      <th>27</th>\n",
       "      <th>28</th>\n",
       "      <th>29</th>\n",
       "      <th>30</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>551472.000000</td>\n",
       "      <td>551472.000000</td>\n",
       "      <td>551472.000000</td>\n",
       "      <td>551472.000000</td>\n",
       "      <td>551472.000000</td>\n",
       "      <td>551472.000000</td>\n",
       "      <td>551472.000000</td>\n",
       "      <td>5.514720e+05</td>\n",
       "      <td>5.514720e+05</td>\n",
       "      <td>5.514720e+05</td>\n",
       "      <td>5.514720e+05</td>\n",
       "      <td>5.514720e+05</td>\n",
       "      <td>551472.000000</td>\n",
       "      <td>5.514720e+05</td>\n",
       "      <td>551472.000000</td>\n",
       "      <td>551472.000000</td>\n",
       "      <td>5.514720e+05</td>\n",
       "      <td>551472.000000</td>\n",
       "      <td>551472.000000</td>\n",
       "      <td>551472.000000</td>\n",
       "      <td>551472.000000</td>\n",
       "      <td>551472.000000</td>\n",
       "      <td>551472.000000</td>\n",
       "      <td>551472.000000</td>\n",
       "      <td>551472.000000</td>\n",
       "      <td>551472.000000</td>\n",
       "      <td>551472.000000</td>\n",
       "      <td>551472.000000</td>\n",
       "      <td>551472.000000</td>\n",
       "      <td>551472.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.050922</td>\n",
       "      <td>0.043485</td>\n",
       "      <td>0.044390</td>\n",
       "      <td>0.037281</td>\n",
       "      <td>0.034938</td>\n",
       "      <td>0.030845</td>\n",
       "      <td>0.035388</td>\n",
       "      <td>4.031147e-02</td>\n",
       "      <td>3.897440e-02</td>\n",
       "      <td>4.033909e-02</td>\n",
       "      <td>3.457703e-02</td>\n",
       "      <td>2.767145e-02</td>\n",
       "      <td>0.029591</td>\n",
       "      <td>3.380846e-02</td>\n",
       "      <td>0.036569</td>\n",
       "      <td>0.036661</td>\n",
       "      <td>3.736359e-02</td>\n",
       "      <td>0.033214</td>\n",
       "      <td>0.029453</td>\n",
       "      <td>0.027493</td>\n",
       "      <td>0.030475</td>\n",
       "      <td>0.035824</td>\n",
       "      <td>0.033340</td>\n",
       "      <td>0.033832</td>\n",
       "      <td>0.027949</td>\n",
       "      <td>0.024122</td>\n",
       "      <td>0.025484</td>\n",
       "      <td>0.026107</td>\n",
       "      <td>0.021173</td>\n",
       "      <td>0.018420</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.110464</td>\n",
       "      <td>0.062425</td>\n",
       "      <td>0.050017</td>\n",
       "      <td>0.033785</td>\n",
       "      <td>0.028101</td>\n",
       "      <td>0.021190</td>\n",
       "      <td>0.021095</td>\n",
       "      <td>2.117326e-02</td>\n",
       "      <td>1.706461e-02</td>\n",
       "      <td>1.679844e-02</td>\n",
       "      <td>1.307059e-02</td>\n",
       "      <td>1.082803e-02</td>\n",
       "      <td>0.011157</td>\n",
       "      <td>1.277828e-02</td>\n",
       "      <td>0.014467</td>\n",
       "      <td>0.015654</td>\n",
       "      <td>1.684666e-02</td>\n",
       "      <td>0.016377</td>\n",
       "      <td>0.015459</td>\n",
       "      <td>0.014807</td>\n",
       "      <td>0.017261</td>\n",
       "      <td>0.021750</td>\n",
       "      <td>0.020812</td>\n",
       "      <td>0.022725</td>\n",
       "      <td>0.019786</td>\n",
       "      <td>0.018146</td>\n",
       "      <td>0.019856</td>\n",
       "      <td>0.021067</td>\n",
       "      <td>0.018363</td>\n",
       "      <td>0.019431</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.376646e-36</td>\n",
       "      <td>1.016562e-31</td>\n",
       "      <td>3.712973e-30</td>\n",
       "      <td>1.695091e-28</td>\n",
       "      <td>3.832159e-32</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>8.590849e-38</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>4.367931e-34</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>0.000215</td>\n",
       "      <td>0.001673</td>\n",
       "      <td>0.004593</td>\n",
       "      <td>0.007384</td>\n",
       "      <td>0.010011</td>\n",
       "      <td>0.011989</td>\n",
       "      <td>0.017092</td>\n",
       "      <td>2.284673e-02</td>\n",
       "      <td>2.551146e-02</td>\n",
       "      <td>2.797258e-02</td>\n",
       "      <td>2.583229e-02</td>\n",
       "      <td>2.107226e-02</td>\n",
       "      <td>0.023779</td>\n",
       "      <td>2.746615e-02</td>\n",
       "      <td>0.028732</td>\n",
       "      <td>0.026773</td>\n",
       "      <td>2.515195e-02</td>\n",
       "      <td>0.019644</td>\n",
       "      <td>0.015915</td>\n",
       "      <td>0.014302</td>\n",
       "      <td>0.014494</td>\n",
       "      <td>0.015479</td>\n",
       "      <td>0.013345</td>\n",
       "      <td>0.011872</td>\n",
       "      <td>0.008919</td>\n",
       "      <td>0.006710</td>\n",
       "      <td>0.007001</td>\n",
       "      <td>0.006579</td>\n",
       "      <td>0.004765</td>\n",
       "      <td>0.003334</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>0.002586</td>\n",
       "      <td>0.011013</td>\n",
       "      <td>0.021321</td>\n",
       "      <td>0.025697</td>\n",
       "      <td>0.027717</td>\n",
       "      <td>0.026817</td>\n",
       "      <td>0.032491</td>\n",
       "      <td>3.799608e-02</td>\n",
       "      <td>3.822447e-02</td>\n",
       "      <td>3.958915e-02</td>\n",
       "      <td>3.495624e-02</td>\n",
       "      <td>2.774311e-02</td>\n",
       "      <td>0.030714</td>\n",
       "      <td>3.610659e-02</td>\n",
       "      <td>0.040904</td>\n",
       "      <td>0.040743</td>\n",
       "      <td>4.212032e-02</td>\n",
       "      <td>0.037674</td>\n",
       "      <td>0.032418</td>\n",
       "      <td>0.030060</td>\n",
       "      <td>0.032963</td>\n",
       "      <td>0.037970</td>\n",
       "      <td>0.035019</td>\n",
       "      <td>0.034246</td>\n",
       "      <td>0.027230</td>\n",
       "      <td>0.022393</td>\n",
       "      <td>0.022458</td>\n",
       "      <td>0.022102</td>\n",
       "      <td>0.016107</td>\n",
       "      <td>0.010624</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>0.035482</td>\n",
       "      <td>0.063040</td>\n",
       "      <td>0.077339</td>\n",
       "      <td>0.066008</td>\n",
       "      <td>0.057902</td>\n",
       "      <td>0.049059</td>\n",
       "      <td>0.052754</td>\n",
       "      <td>5.699276e-02</td>\n",
       "      <td>5.313037e-02</td>\n",
       "      <td>5.349386e-02</td>\n",
       "      <td>4.433072e-02</td>\n",
       "      <td>3.545784e-02</td>\n",
       "      <td>0.037323</td>\n",
       "      <td>4.314112e-02</td>\n",
       "      <td>0.047498</td>\n",
       "      <td>0.048844</td>\n",
       "      <td>5.049463e-02</td>\n",
       "      <td>0.046278</td>\n",
       "      <td>0.042161</td>\n",
       "      <td>0.040170</td>\n",
       "      <td>0.046187</td>\n",
       "      <td>0.054441</td>\n",
       "      <td>0.051996</td>\n",
       "      <td>0.053518</td>\n",
       "      <td>0.044612</td>\n",
       "      <td>0.038878</td>\n",
       "      <td>0.041585</td>\n",
       "      <td>0.042907</td>\n",
       "      <td>0.034881</td>\n",
       "      <td>0.029869</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>0.999872</td>\n",
       "      <td>0.998519</td>\n",
       "      <td>0.999998</td>\n",
       "      <td>0.220135</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.302767</td>\n",
       "      <td>0.261616</td>\n",
       "      <td>5.057332e-01</td>\n",
       "      <td>5.086496e-01</td>\n",
       "      <td>3.373353e-01</td>\n",
       "      <td>2.560698e-01</td>\n",
       "      <td>5.587589e-01</td>\n",
       "      <td>0.712937</td>\n",
       "      <td>3.592974e-01</td>\n",
       "      <td>0.706429</td>\n",
       "      <td>0.311662</td>\n",
       "      <td>6.472700e-01</td>\n",
       "      <td>0.106529</td>\n",
       "      <td>0.513319</td>\n",
       "      <td>0.085274</td>\n",
       "      <td>0.145688</td>\n",
       "      <td>0.913767</td>\n",
       "      <td>0.517235</td>\n",
       "      <td>0.145636</td>\n",
       "      <td>0.859147</td>\n",
       "      <td>0.459230</td>\n",
       "      <td>0.533557</td>\n",
       "      <td>0.945968</td>\n",
       "      <td>0.176657</td>\n",
       "      <td>0.879738</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                  1              2              3              4   \\\n",
       "count  551472.000000  551472.000000  551472.000000  551472.000000   \n",
       "mean        0.050922       0.043485       0.044390       0.037281   \n",
       "std         0.110464       0.062425       0.050017       0.033785   \n",
       "min         0.000000       0.000000       0.000000       0.000000   \n",
       "25%         0.000215       0.001673       0.004593       0.007384   \n",
       "50%         0.002586       0.011013       0.021321       0.025697   \n",
       "75%         0.035482       0.063040       0.077339       0.066008   \n",
       "max         0.999872       0.998519       0.999998       0.220135   \n",
       "\n",
       "                  5              6              7             8   \\\n",
       "count  551472.000000  551472.000000  551472.000000  5.514720e+05   \n",
       "mean        0.034938       0.030845       0.035388  4.031147e-02   \n",
       "std         0.028101       0.021190       0.021095  2.117326e-02   \n",
       "min         0.000000       0.000000       0.000000  1.376646e-36   \n",
       "25%         0.010011       0.011989       0.017092  2.284673e-02   \n",
       "50%         0.027717       0.026817       0.032491  3.799608e-02   \n",
       "75%         0.057902       0.049059       0.052754  5.699276e-02   \n",
       "max         1.000000       0.302767       0.261616  5.057332e-01   \n",
       "\n",
       "                 9             10            11            12             13  \\\n",
       "count  5.514720e+05  5.514720e+05  5.514720e+05  5.514720e+05  551472.000000   \n",
       "mean   3.897440e-02  4.033909e-02  3.457703e-02  2.767145e-02       0.029591   \n",
       "std    1.706461e-02  1.679844e-02  1.307059e-02  1.082803e-02       0.011157   \n",
       "min    1.016562e-31  3.712973e-30  1.695091e-28  3.832159e-32       0.000000   \n",
       "25%    2.551146e-02  2.797258e-02  2.583229e-02  2.107226e-02       0.023779   \n",
       "50%    3.822447e-02  3.958915e-02  3.495624e-02  2.774311e-02       0.030714   \n",
       "75%    5.313037e-02  5.349386e-02  4.433072e-02  3.545784e-02       0.037323   \n",
       "max    5.086496e-01  3.373353e-01  2.560698e-01  5.587589e-01       0.712937   \n",
       "\n",
       "                 14             15             16            17  \\\n",
       "count  5.514720e+05  551472.000000  551472.000000  5.514720e+05   \n",
       "mean   3.380846e-02       0.036569       0.036661  3.736359e-02   \n",
       "std    1.277828e-02       0.014467       0.015654  1.684666e-02   \n",
       "min    8.590849e-38       0.000000       0.000000  4.367931e-34   \n",
       "25%    2.746615e-02       0.028732       0.026773  2.515195e-02   \n",
       "50%    3.610659e-02       0.040904       0.040743  4.212032e-02   \n",
       "75%    4.314112e-02       0.047498       0.048844  5.049463e-02   \n",
       "max    3.592974e-01       0.706429       0.311662  6.472700e-01   \n",
       "\n",
       "                  18             19             20             21  \\\n",
       "count  551472.000000  551472.000000  551472.000000  551472.000000   \n",
       "mean        0.033214       0.029453       0.027493       0.030475   \n",
       "std         0.016377       0.015459       0.014807       0.017261   \n",
       "min         0.000000       0.000000       0.000000       0.000000   \n",
       "25%         0.019644       0.015915       0.014302       0.014494   \n",
       "50%         0.037674       0.032418       0.030060       0.032963   \n",
       "75%         0.046278       0.042161       0.040170       0.046187   \n",
       "max         0.106529       0.513319       0.085274       0.145688   \n",
       "\n",
       "                  22             23             24             25  \\\n",
       "count  551472.000000  551472.000000  551472.000000  551472.000000   \n",
       "mean        0.035824       0.033340       0.033832       0.027949   \n",
       "std         0.021750       0.020812       0.022725       0.019786   \n",
       "min         0.000000       0.000000       0.000000       0.000000   \n",
       "25%         0.015479       0.013345       0.011872       0.008919   \n",
       "50%         0.037970       0.035019       0.034246       0.027230   \n",
       "75%         0.054441       0.051996       0.053518       0.044612   \n",
       "max         0.913767       0.517235       0.145636       0.859147   \n",
       "\n",
       "                  26             27             28             29  \\\n",
       "count  551472.000000  551472.000000  551472.000000  551472.000000   \n",
       "mean        0.024122       0.025484       0.026107       0.021173   \n",
       "std         0.018146       0.019856       0.021067       0.018363   \n",
       "min         0.000000       0.000000       0.000000       0.000000   \n",
       "25%         0.006710       0.007001       0.006579       0.004765   \n",
       "50%         0.022393       0.022458       0.022102       0.016107   \n",
       "75%         0.038878       0.041585       0.042907       0.034881   \n",
       "max         0.459230       0.533557       0.945968       0.176657   \n",
       "\n",
       "                  30  \n",
       "count  551472.000000  \n",
       "mean        0.018420  \n",
       "std         0.019431  \n",
       "min         0.000000  \n",
       "25%         0.003334  \n",
       "50%         0.010624  \n",
       "75%         0.029869  \n",
       "max         0.879738  "
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test_sub.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "9ca6ac1e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 1min 46s, sys: 1.4 s, total: 1min 47s\n",
      "Wall time: 1min 51s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "X_test_sub.to_csv(\n",
    "    '3.1-nn-one-layer-32.csv.gz',\n",
    "    compression='gzip',\n",
    "    index=False,\n",
    "    header=False\n",
    ")"
   ]
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "31f2aee4e71d21fbe5cf8b01ff0e069b9275f58929596ceb00d14d90e3e16cd6"
  },
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}