Augmented LightGBM model .ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Augmented answer generation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import StratifiedKFold\n",
    "from skopt.space import Real, Integer\n",
    "from skopt.utils import use_named_args\n",
    "import itertools\n",
    "from sklearn.metrics import roc_auc_score\n",
    "from skopt import gp_minimize\n",
    "from fastai import *\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import pandas_profiling\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import lightgbm as lgb"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "TRAIN = 'data/train.csv'\n",
    "train = pd.read_csv(TRAIN)\n",
    "X = train.drop(['ID_code', 'target'], axis=1)\n",
    "y = train.target"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pickle\n",
    "def save_model(models, filename):\n",
    "    with open(filename, 'wb') as handle:\n",
    "        pickle.dump(models, handle)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_preds(X, y, params):\n",
    "    skf = StratifiedKFold(n_splits=5)\n",
    "    models = []\n",
    "    for train_idx, test_idx in skf.split(X, y):\n",
    "            trn_data = lgb.Dataset(X[train_idx], label=y[train_idx])\n",
    "            test_data = lgb.Dataset(X[test_idx], label=y[test_idx])\n",
    "            model_lgb     = lgb.train(params, trn_data, 100000, valid_sets = [trn_data, test_data], verbose_eval=5000, early_stopping_rounds = 4000)\n",
    "            models.append(model_lgb)\n",
    "            auc = roc_auc_score(y[test_idx], model_lgb.predict(X[test_idx]))\n",
    "            print(f\"This AUC: {auc}\")\n",
    "    return models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "params = {\n",
    "    'bagging_freq': 5,\n",
    "    'bagging_fraction': 0.335,\n",
    "    'boost_from_average':'false',\n",
    "    'boost': 'gbdt',\n",
    "    'feature_fraction': 0.041,\n",
    "    'learning_rate': 0.0053,\n",
    "    'max_depth': -1,\n",
    "    'metric':'auc',\n",
    "    'min_data_in_leaf': 80,\n",
    "    'min_sum_hessian_in_leaf': 10.0,\n",
    "    'num_leaves': 13,\n",
    "    'num_threads': 8,\n",
    "    'tree_learner': 'serial',\n",
    "    'objective': 'binary', \n",
    "    'verbosity': -1,\n",
    "    'gpu_platform_id': -1,\n",
    "}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Random super-sampling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "from imblearn.over_sampling import RandomOverSampler\n",
    "ros = RandomOverSampler(random_state=0)\n",
    "X_resampled, y_resampled = ros.fit_resample(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training until validation scores don't improve for 4000 rounds.\n",
      "[5000]\ttraining's auc: 0.920294\tvalid_1's auc: 0.915505\n",
      "[10000]\ttraining's auc: 0.934396\tvalid_1's auc: 0.927559\n",
      "[15000]\ttraining's auc: 0.945706\tvalid_1's auc: 0.937056\n",
      "[20000]\ttraining's auc: 0.95526\tvalid_1's auc: 0.945119\n",
      "[25000]\ttraining's auc: 0.963304\tvalid_1's auc: 0.952058\n",
      "[30000]\ttraining's auc: 0.970102\tvalid_1's auc: 0.958076\n",
      "[35000]\ttraining's auc: 0.975699\tvalid_1's auc: 0.963216\n",
      "[40000]\ttraining's auc: 0.980349\tvalid_1's auc: 0.967655\n",
      "[45000]\ttraining's auc: 0.98412\tvalid_1's auc: 0.971439\n",
      "[50000]\ttraining's auc: 0.987216\tvalid_1's auc: 0.974739\n",
      "[55000]\ttraining's auc: 0.989739\tvalid_1's auc: 0.977526\n",
      "[60000]\ttraining's auc: 0.991776\tvalid_1's auc: 0.97994\n",
      "[65000]\ttraining's auc: 0.993425\tvalid_1's auc: 0.982034\n",
      "[70000]\ttraining's auc: 0.994753\tvalid_1's auc: 0.983844\n",
      "[75000]\ttraining's auc: 0.995824\tvalid_1's auc: 0.98542\n",
      "[80000]\ttraining's auc: 0.996682\tvalid_1's auc: 0.986787\n",
      "[85000]\ttraining's auc: 0.997377\tvalid_1's auc: 0.987995\n",
      "[90000]\ttraining's auc: 0.997928\tvalid_1's auc: 0.98905\n",
      "[95000]\ttraining's auc: 0.998372\tvalid_1's auc: 0.989972\n",
      "[100000]\ttraining's auc: 0.998722\tvalid_1's auc: 0.990788\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[100000]\ttraining's auc: 0.998722\tvalid_1's auc: 0.990788\n",
      "This AUC: 0.9907883014829104\n",
      "Training until validation scores don't improve for 4000 rounds.\n",
      "[5000]\ttraining's auc: 0.919945\tvalid_1's auc: 0.915983\n",
      "[10000]\ttraining's auc: 0.933928\tvalid_1's auc: 0.928041\n",
      "[15000]\ttraining's auc: 0.945235\tvalid_1's auc: 0.93762\n",
      "[20000]\ttraining's auc: 0.954724\tvalid_1's auc: 0.945734\n",
      "[25000]\ttraining's auc: 0.962909\tvalid_1's auc: 0.95283\n",
      "[30000]\ttraining's auc: 0.969708\tvalid_1's auc: 0.958919\n",
      "[35000]\ttraining's auc: 0.975348\tvalid_1's auc: 0.964088\n",
      "[40000]\ttraining's auc: 0.980063\tvalid_1's auc: 0.968581\n",
      "[45000]\ttraining's auc: 0.983916\tvalid_1's auc: 0.972396\n",
      "[50000]\ttraining's auc: 0.987069\tvalid_1's auc: 0.975661\n",
      "[55000]\ttraining's auc: 0.989628\tvalid_1's auc: 0.978398\n",
      "[60000]\ttraining's auc: 0.991701\tvalid_1's auc: 0.980801\n",
      "[65000]\ttraining's auc: 0.993382\tvalid_1's auc: 0.982862\n",
      "[70000]\ttraining's auc: 0.994743\tvalid_1's auc: 0.984638\n",
      "[75000]\ttraining's auc: 0.995834\tvalid_1's auc: 0.986168\n",
      "[80000]\ttraining's auc: 0.996707\tvalid_1's auc: 0.987528\n",
      "[85000]\ttraining's auc: 0.997411\tvalid_1's auc: 0.988704\n",
      "[90000]\ttraining's auc: 0.997967\tvalid_1's auc: 0.989738\n",
      "[95000]\ttraining's auc: 0.998415\tvalid_1's auc: 0.990642\n",
      "[100000]\ttraining's auc: 0.998768\tvalid_1's auc: 0.99144\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[100000]\ttraining's auc: 0.998768\tvalid_1's auc: 0.99144\n",
      "This AUC: 0.9914399328073029\n",
      "Training until validation scores don't improve for 4000 rounds.\n",
      "[5000]\ttraining's auc: 0.920186\tvalid_1's auc: 0.914391\n",
      "[10000]\ttraining's auc: 0.934168\tvalid_1's auc: 0.926566\n",
      "[15000]\ttraining's auc: 0.945468\tvalid_1's auc: 0.936165\n",
      "[20000]\ttraining's auc: 0.954974\tvalid_1's auc: 0.944258\n",
      "[25000]\ttraining's auc: 0.9631\tvalid_1's auc: 0.951408\n",
      "[30000]\ttraining's auc: 0.969909\tvalid_1's auc: 0.957469\n",
      "[35000]\ttraining's auc: 0.975533\tvalid_1's auc: 0.962663\n",
      "[40000]\ttraining's auc: 0.980198\tvalid_1's auc: 0.967155\n",
      "[45000]\ttraining's auc: 0.984027\tvalid_1's auc: 0.971035\n",
      "[50000]\ttraining's auc: 0.987163\tvalid_1's auc: 0.974401\n",
      "[55000]\ttraining's auc: 0.989719\tvalid_1's auc: 0.977277\n",
      "[60000]\ttraining's auc: 0.99178\tvalid_1's auc: 0.979737\n",
      "[65000]\ttraining's auc: 0.993442\tvalid_1's auc: 0.981835\n",
      "[70000]\ttraining's auc: 0.994794\tvalid_1's auc: 0.983693\n",
      "[75000]\ttraining's auc: 0.995879\tvalid_1's auc: 0.985279\n",
      "[80000]\ttraining's auc: 0.996747\tvalid_1's auc: 0.986697\n",
      "[85000]\ttraining's auc: 0.997445\tvalid_1's auc: 0.987914\n",
      "[90000]\ttraining's auc: 0.997992\tvalid_1's auc: 0.988973\n",
      "[95000]\ttraining's auc: 0.998434\tvalid_1's auc: 0.989913\n",
      "[100000]\ttraining's auc: 0.998781\tvalid_1's auc: 0.990741\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[100000]\ttraining's auc: 0.998781\tvalid_1's auc: 0.990741\n",
      "This AUC: 0.9907405826719247\n",
      "Training until validation scores don't improve for 4000 rounds.\n",
      "[5000]\ttraining's auc: 0.919849\tvalid_1's auc: 0.916162\n",
      "[10000]\ttraining's auc: 0.933877\tvalid_1's auc: 0.928221\n",
      "[15000]\ttraining's auc: 0.945165\tvalid_1's auc: 0.937656\n",
      "[20000]\ttraining's auc: 0.954711\tvalid_1's auc: 0.945665\n",
      "[25000]\ttraining's auc: 0.962848\tvalid_1's auc: 0.952625\n",
      "[30000]\ttraining's auc: 0.969686\tvalid_1's auc: 0.958647\n",
      "[35000]\ttraining's auc: 0.975335\tvalid_1's auc: 0.96376\n",
      "[40000]\ttraining's auc: 0.98001\tvalid_1's auc: 0.96815\n",
      "[45000]\ttraining's auc: 0.983865\tvalid_1's auc: 0.97197\n",
      "[50000]\ttraining's auc: 0.987031\tvalid_1's auc: 0.975225\n",
      "[55000]\ttraining's auc: 0.989583\tvalid_1's auc: 0.977998\n",
      "[60000]\ttraining's auc: 0.991672\tvalid_1's auc: 0.980407\n",
      "[65000]\ttraining's auc: 0.993348\tvalid_1's auc: 0.982466\n",
      "[70000]\ttraining's auc: 0.994716\tvalid_1's auc: 0.98427\n",
      "[75000]\ttraining's auc: 0.995809\tvalid_1's auc: 0.985825\n",
      "[80000]\ttraining's auc: 0.996691\tvalid_1's auc: 0.987171\n",
      "[85000]\ttraining's auc: 0.997381\tvalid_1's auc: 0.988328\n",
      "[90000]\ttraining's auc: 0.997939\tvalid_1's auc: 0.98937\n",
      "[95000]\ttraining's auc: 0.998388\tvalid_1's auc: 0.990291\n",
      "[100000]\ttraining's auc: 0.998741\tvalid_1's auc: 0.991082\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[100000]\ttraining's auc: 0.998741\tvalid_1's auc: 0.991082\n",
      "This AUC: 0.9910821959330751\n",
      "Training until validation scores don't improve for 4000 rounds.\n",
      "[5000]\ttraining's auc: 0.920198\tvalid_1's auc: 0.914125\n",
      "[10000]\ttraining's auc: 0.934071\tvalid_1's auc: 0.926558\n",
      "[15000]\ttraining's auc: 0.94531\tvalid_1's auc: 0.93637\n",
      "[20000]\ttraining's auc: 0.954808\tvalid_1's auc: 0.944696\n",
      "[25000]\ttraining's auc: 0.962891\tvalid_1's auc: 0.951924\n",
      "[30000]\ttraining's auc: 0.969707\tvalid_1's auc: 0.958163\n",
      "[35000]\ttraining's auc: 0.975385\tvalid_1's auc: 0.963483\n",
      "[40000]\ttraining's auc: 0.980053\tvalid_1's auc: 0.96804\n",
      "[45000]\ttraining's auc: 0.983922\tvalid_1's auc: 0.971934\n",
      "[50000]\ttraining's auc: 0.987064\tvalid_1's auc: 0.97526\n",
      "[55000]\ttraining's auc: 0.989615\tvalid_1's auc: 0.978106\n",
      "[60000]\ttraining's auc: 0.991707\tvalid_1's auc: 0.980546\n",
      "[65000]\ttraining's auc: 0.993387\tvalid_1's auc: 0.982672\n",
      "[70000]\ttraining's auc: 0.994744\tvalid_1's auc: 0.9845\n",
      "[75000]\ttraining's auc: 0.995841\tvalid_1's auc: 0.986075\n",
      "[80000]\ttraining's auc: 0.996701\tvalid_1's auc: 0.98742\n",
      "[85000]\ttraining's auc: 0.997401\tvalid_1's auc: 0.988619\n",
      "[90000]\ttraining's auc: 0.997956\tvalid_1's auc: 0.989643\n",
      "[95000]\ttraining's auc: 0.9984\tvalid_1's auc: 0.990555\n",
      "[100000]\ttraining's auc: 0.998749\tvalid_1's auc: 0.991335\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[100000]\ttraining's auc: 0.998749\tvalid_1's auc: 0.991335\n",
      "This AUC: 0.991335099544216\n"
     ]
    }
   ],
   "source": [
    "random_trained_models = get_preds(X_resampled, y_resampled, params)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "save_model(random_trained_models, 'random_aug_lgbm_models_cv10.m')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### SMOTE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "from imblearn.over_sampling import SMOTE\n",
    "smote = SMOTE(random_state=0)\n",
    "X_resampled, y_resampled = smote.fit_resample(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training until validation scores don't improve for 4000 rounds.\n",
      "[5000]\ttraining's auc: 0.97333\tvalid_1's auc: 0.813928\n",
      "[10000]\ttraining's auc: 0.980199\tvalid_1's auc: 0.834526\n",
      "[15000]\ttraining's auc: 0.983753\tvalid_1's auc: 0.846639\n",
      "[20000]\ttraining's auc: 0.986295\tvalid_1's auc: 0.855567\n",
      "[25000]\ttraining's auc: 0.988356\tvalid_1's auc: 0.862391\n",
      "[30000]\ttraining's auc: 0.99013\tvalid_1's auc: 0.86792\n",
      "[35000]\ttraining's auc: 0.991656\tvalid_1's auc: 0.873065\n",
      "[40000]\ttraining's auc: 0.992983\tvalid_1's auc: 0.877587\n",
      "[45000]\ttraining's auc: 0.994121\tvalid_1's auc: 0.881634\n",
      "[50000]\ttraining's auc: 0.995098\tvalid_1's auc: 0.885424\n",
      "[55000]\ttraining's auc: 0.995928\tvalid_1's auc: 0.888881\n",
      "[60000]\ttraining's auc: 0.996637\tvalid_1's auc: 0.892138\n",
      "[65000]\ttraining's auc: 0.997233\tvalid_1's auc: 0.895116\n",
      "[70000]\ttraining's auc: 0.997739\tvalid_1's auc: 0.897938\n",
      "[75000]\ttraining's auc: 0.99816\tvalid_1's auc: 0.900547\n",
      "[80000]\ttraining's auc: 0.998507\tvalid_1's auc: 0.902909\n",
      "[85000]\ttraining's auc: 0.998796\tvalid_1's auc: 0.90514\n",
      "[90000]\ttraining's auc: 0.999033\tvalid_1's auc: 0.90719\n",
      "[95000]\ttraining's auc: 0.999228\tvalid_1's auc: 0.9091\n",
      "[100000]\ttraining's auc: 0.99939\tvalid_1's auc: 0.910971\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[100000]\ttraining's auc: 0.99939\tvalid_1's auc: 0.910971\n",
      "This AUC: 0.9109706350063963\n",
      "Training until validation scores don't improve for 4000 rounds.\n",
      "[5000]\ttraining's auc: 0.94673\tvalid_1's auc: 0.971116\n",
      "[10000]\ttraining's auc: 0.95878\tvalid_1's auc: 0.976418\n",
      "[15000]\ttraining's auc: 0.964781\tvalid_1's auc: 0.978799\n",
      "[20000]\ttraining's auc: 0.969164\tvalid_1's auc: 0.980455\n",
      "[25000]\ttraining's auc: 0.97286\tvalid_1's auc: 0.981841\n",
      "[30000]\ttraining's auc: 0.976093\tvalid_1's auc: 0.983063\n",
      "[35000]\ttraining's auc: 0.978975\tvalid_1's auc: 0.984207\n",
      "[40000]\ttraining's auc: 0.981574\tvalid_1's auc: 0.985276\n",
      "[45000]\ttraining's auc: 0.983896\tvalid_1's auc: 0.986241\n",
      "[50000]\ttraining's auc: 0.985971\tvalid_1's auc: 0.987132\n",
      "[55000]\ttraining's auc: 0.987809\tvalid_1's auc: 0.987928\n",
      "[60000]\ttraining's auc: 0.989434\tvalid_1's auc: 0.988682\n",
      "[65000]\ttraining's auc: 0.990882\tvalid_1's auc: 0.989362\n",
      "[70000]\ttraining's auc: 0.992159\tvalid_1's auc: 0.98999\n",
      "[75000]\ttraining's auc: 0.993272\tvalid_1's auc: 0.99056\n",
      "[80000]\ttraining's auc: 0.994247\tvalid_1's auc: 0.991102\n",
      "[85000]\ttraining's auc: 0.995088\tvalid_1's auc: 0.991585\n",
      "[90000]\ttraining's auc: 0.995821\tvalid_1's auc: 0.992041\n",
      "[95000]\ttraining's auc: 0.996452\tvalid_1's auc: 0.992447\n",
      "[100000]\ttraining's auc: 0.996998\tvalid_1's auc: 0.992832\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[100000]\ttraining's auc: 0.996998\tvalid_1's auc: 0.992832\n",
      "This AUC: 0.992832151984188\n",
      "Training until validation scores don't improve for 4000 rounds.\n",
      "[5000]\ttraining's auc: 0.946967\tvalid_1's auc: 0.970979\n",
      "[10000]\ttraining's auc: 0.958927\tvalid_1's auc: 0.97636\n",
      "[15000]\ttraining's auc: 0.964897\tvalid_1's auc: 0.978781\n",
      "[20000]\ttraining's auc: 0.969252\tvalid_1's auc: 0.980445\n",
      "[25000]\ttraining's auc: 0.972909\tvalid_1's auc: 0.981854\n",
      "[30000]\ttraining's auc: 0.976142\tvalid_1's auc: 0.9831\n",
      "[35000]\ttraining's auc: 0.97901\tvalid_1's auc: 0.984241\n",
      "[40000]\ttraining's auc: 0.981586\tvalid_1's auc: 0.985299\n",
      "[45000]\ttraining's auc: 0.983924\tvalid_1's auc: 0.986246\n",
      "[50000]\ttraining's auc: 0.98602\tvalid_1's auc: 0.987118\n",
      "[55000]\ttraining's auc: 0.987882\tvalid_1's auc: 0.987913\n",
      "[60000]\ttraining's auc: 0.989523\tvalid_1's auc: 0.988659\n",
      "[65000]\ttraining's auc: 0.990968\tvalid_1's auc: 0.989323\n",
      "[70000]\ttraining's auc: 0.992238\tvalid_1's auc: 0.989945\n",
      "[75000]\ttraining's auc: 0.993343\tvalid_1's auc: 0.99051\n",
      "[80000]\ttraining's auc: 0.994319\tvalid_1's auc: 0.991018\n",
      "[85000]\ttraining's auc: 0.99516\tvalid_1's auc: 0.991489\n",
      "[90000]\ttraining's auc: 0.995884\tvalid_1's auc: 0.991916\n",
      "[95000]\ttraining's auc: 0.996514\tvalid_1's auc: 0.992307\n",
      "[100000]\ttraining's auc: 0.997055\tvalid_1's auc: 0.992682\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[100000]\ttraining's auc: 0.997055\tvalid_1's auc: 0.992682\n",
      "This AUC: 0.9926824951543397\n",
      "Training until validation scores don't improve for 4000 rounds.\n",
      "[5000]\ttraining's auc: 0.946588\tvalid_1's auc: 0.972507\n",
      "[10000]\ttraining's auc: 0.958604\tvalid_1's auc: 0.977677\n",
      "[15000]\ttraining's auc: 0.964643\tvalid_1's auc: 0.980019\n",
      "[20000]\ttraining's auc: 0.969063\tvalid_1's auc: 0.981609\n",
      "[25000]\ttraining's auc: 0.972763\tvalid_1's auc: 0.982941\n",
      "[30000]\ttraining's auc: 0.975979\tvalid_1's auc: 0.984125\n",
      "[35000]\ttraining's auc: 0.978891\tvalid_1's auc: 0.985235\n",
      "[40000]\ttraining's auc: 0.981513\tvalid_1's auc: 0.986229\n",
      "[45000]\ttraining's auc: 0.983855\tvalid_1's auc: 0.987136\n",
      "[50000]\ttraining's auc: 0.985947\tvalid_1's auc: 0.987967\n",
      "[55000]\ttraining's auc: 0.987801\tvalid_1's auc: 0.988718\n",
      "[60000]\ttraining's auc: 0.989455\tvalid_1's auc: 0.989408\n",
      "[65000]\ttraining's auc: 0.990891\tvalid_1's auc: 0.990052\n",
      "[70000]\ttraining's auc: 0.992157\tvalid_1's auc: 0.990636\n",
      "[75000]\ttraining's auc: 0.993265\tvalid_1's auc: 0.991163\n",
      "[80000]\ttraining's auc: 0.99425\tvalid_1's auc: 0.991654\n",
      "[85000]\ttraining's auc: 0.995096\tvalid_1's auc: 0.992102\n",
      "[90000]\ttraining's auc: 0.995824\tvalid_1's auc: 0.992515\n",
      "[95000]\ttraining's auc: 0.996459\tvalid_1's auc: 0.992903\n",
      "[100000]\ttraining's auc: 0.997003\tvalid_1's auc: 0.993247\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[100000]\ttraining's auc: 0.997003\tvalid_1's auc: 0.993247\n",
      "This AUC: 0.9932467762801952\n",
      "Training until validation scores don't improve for 4000 rounds.\n",
      "[5000]\ttraining's auc: 0.946622\tvalid_1's auc: 0.972287\n",
      "[10000]\ttraining's auc: 0.958657\tvalid_1's auc: 0.977449\n",
      "[15000]\ttraining's auc: 0.964708\tvalid_1's auc: 0.979729\n",
      "[20000]\ttraining's auc: 0.969101\tvalid_1's auc: 0.981321\n",
      "[25000]\ttraining's auc: 0.972781\tvalid_1's auc: 0.98266\n",
      "[30000]\ttraining's auc: 0.976\tvalid_1's auc: 0.983886\n",
      "[35000]\ttraining's auc: 0.978901\tvalid_1's auc: 0.985008\n",
      "[40000]\ttraining's auc: 0.981513\tvalid_1's auc: 0.986027\n",
      "[45000]\ttraining's auc: 0.983847\tvalid_1's auc: 0.986967\n",
      "[50000]\ttraining's auc: 0.985939\tvalid_1's auc: 0.987815\n",
      "[55000]\ttraining's auc: 0.987794\tvalid_1's auc: 0.988606\n",
      "[60000]\ttraining's auc: 0.989456\tvalid_1's auc: 0.98934\n",
      "[65000]\ttraining's auc: 0.9909\tvalid_1's auc: 0.989994\n",
      "[70000]\ttraining's auc: 0.992171\tvalid_1's auc: 0.990605\n",
      "[75000]\ttraining's auc: 0.993282\tvalid_1's auc: 0.991156\n",
      "[80000]\ttraining's auc: 0.994255\tvalid_1's auc: 0.991659\n",
      "[85000]\ttraining's auc: 0.995097\tvalid_1's auc: 0.992115\n",
      "[90000]\ttraining's auc: 0.995821\tvalid_1's auc: 0.992531\n",
      "[95000]\ttraining's auc: 0.996453\tvalid_1's auc: 0.99292\n",
      "[100000]\ttraining's auc: 0.997005\tvalid_1's auc: 0.993285\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[100000]\ttraining's auc: 0.997005\tvalid_1's auc: 0.993285\n",
      "This AUC: 0.9932846122900099\n"
     ]
    }
   ],
   "source": [
    "smote_trained_models = get_preds(X_resampled, y_resampled, params)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "save_model(smote_trained_models, 'smote_aug_lgbm_models_cv10.m')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### ADASYN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "from imblearn.over_sampling import  ADASYN\n",
    "adasyn = ADASYN(random_state=0)\n",
    "X_resampled, y_resampled = adasyn.fit_resample(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training until validation scores don't improve for 4000 rounds.\n",
      "[5000]\ttraining's auc: 0.972971\tvalid_1's auc: 0.819399\n",
      "[10000]\ttraining's auc: 0.979884\tvalid_1's auc: 0.839541\n",
      "[15000]\ttraining's auc: 0.983455\tvalid_1's auc: 0.850974\n",
      "[20000]\ttraining's auc: 0.986019\tvalid_1's auc: 0.859532\n",
      "[25000]\ttraining's auc: 0.988089\tvalid_1's auc: 0.866061\n",
      "[30000]\ttraining's auc: 0.989854\tvalid_1's auc: 0.871383\n",
      "[35000]\ttraining's auc: 0.991397\tvalid_1's auc: 0.876201\n",
      "[40000]\ttraining's auc: 0.99274\tvalid_1's auc: 0.880395\n",
      "[45000]\ttraining's auc: 0.993908\tvalid_1's auc: 0.884331\n",
      "[50000]\ttraining's auc: 0.994912\tvalid_1's auc: 0.88792\n",
      "[55000]\ttraining's auc: 0.995769\tvalid_1's auc: 0.891184\n",
      "[60000]\ttraining's auc: 0.9965\tvalid_1's auc: 0.894353\n",
      "[65000]\ttraining's auc: 0.997123\tvalid_1's auc: 0.897243\n",
      "[70000]\ttraining's auc: 0.997651\tvalid_1's auc: 0.89988\n",
      "[75000]\ttraining's auc: 0.99809\tvalid_1's auc: 0.902408\n",
      "[80000]\ttraining's auc: 0.998455\tvalid_1's auc: 0.904735\n",
      "[85000]\ttraining's auc: 0.998755\tvalid_1's auc: 0.90685\n",
      "[90000]\ttraining's auc: 0.999003\tvalid_1's auc: 0.908831\n",
      "[95000]\ttraining's auc: 0.999207\tvalid_1's auc: 0.910745\n",
      "[100000]\ttraining's auc: 0.999372\tvalid_1's auc: 0.912461\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[100000]\ttraining's auc: 0.999372\tvalid_1's auc: 0.912461\n",
      "This AUC: 0.9124605738481447\n",
      "Training until validation scores don't improve for 4000 rounds.\n",
      "[5000]\ttraining's auc: 0.947209\tvalid_1's auc: 0.970301\n",
      "[10000]\ttraining's auc: 0.9591\tvalid_1's auc: 0.97547\n",
      "[15000]\ttraining's auc: 0.965054\tvalid_1's auc: 0.977656\n",
      "[20000]\ttraining's auc: 0.96942\tvalid_1's auc: 0.979102\n",
      "[25000]\ttraining's auc: 0.973089\tvalid_1's auc: 0.980276\n",
      "[30000]\ttraining's auc: 0.976299\tvalid_1's auc: 0.981333\n",
      "[35000]\ttraining's auc: 0.979195\tvalid_1's auc: 0.982296\n",
      "[40000]\ttraining's auc: 0.981769\tvalid_1's auc: 0.983209\n",
      "[45000]\ttraining's auc: 0.984072\tvalid_1's auc: 0.984068\n",
      "[50000]\ttraining's auc: 0.986137\tvalid_1's auc: 0.984878\n",
      "[55000]\ttraining's auc: 0.987965\tvalid_1's auc: 0.985617\n",
      "[60000]\ttraining's auc: 0.989585\tvalid_1's auc: 0.986314\n",
      "[65000]\ttraining's auc: 0.990999\tvalid_1's auc: 0.986959\n",
      "[70000]\ttraining's auc: 0.992248\tvalid_1's auc: 0.987559\n",
      "[75000]\ttraining's auc: 0.993335\tvalid_1's auc: 0.988121\n",
      "[80000]\ttraining's auc: 0.994291\tvalid_1's auc: 0.988653\n",
      "[85000]\ttraining's auc: 0.995115\tvalid_1's auc: 0.989155\n",
      "[90000]\ttraining's auc: 0.995826\tvalid_1's auc: 0.989615\n",
      "[95000]\ttraining's auc: 0.996444\tvalid_1's auc: 0.990052\n",
      "[100000]\ttraining's auc: 0.996977\tvalid_1's auc: 0.990468\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[100000]\ttraining's auc: 0.996977\tvalid_1's auc: 0.990468\n",
      "This AUC: 0.9904681241362125\n",
      "Training until validation scores don't improve for 4000 rounds.\n",
      "[5000]\ttraining's auc: 0.947621\tvalid_1's auc: 0.969639\n",
      "[10000]\ttraining's auc: 0.959467\tvalid_1's auc: 0.974542\n",
      "[15000]\ttraining's auc: 0.965445\tvalid_1's auc: 0.97668\n",
      "[20000]\ttraining's auc: 0.969777\tvalid_1's auc: 0.978061\n",
      "[25000]\ttraining's auc: 0.973432\tvalid_1's auc: 0.979192\n",
      "[30000]\ttraining's auc: 0.976628\tvalid_1's auc: 0.98025\n",
      "[35000]\ttraining's auc: 0.97949\tvalid_1's auc: 0.981232\n",
      "[40000]\ttraining's auc: 0.982046\tvalid_1's auc: 0.982167\n",
      "[45000]\ttraining's auc: 0.984331\tvalid_1's auc: 0.983044\n",
      "[50000]\ttraining's auc: 0.986375\tvalid_1's auc: 0.983855\n",
      "[55000]\ttraining's auc: 0.988169\tvalid_1's auc: 0.984581\n",
      "[60000]\ttraining's auc: 0.989759\tvalid_1's auc: 0.985284\n",
      "[65000]\ttraining's auc: 0.991165\tvalid_1's auc: 0.985952\n",
      "[70000]\ttraining's auc: 0.992394\tvalid_1's auc: 0.986572\n",
      "[75000]\ttraining's auc: 0.993449\tvalid_1's auc: 0.987137\n",
      "[80000]\ttraining's auc: 0.994377\tvalid_1's auc: 0.987698\n",
      "[85000]\ttraining's auc: 0.995185\tvalid_1's auc: 0.988207\n",
      "[90000]\ttraining's auc: 0.995893\tvalid_1's auc: 0.988686\n",
      "[95000]\ttraining's auc: 0.996501\tvalid_1's auc: 0.989145\n",
      "[100000]\ttraining's auc: 0.997024\tvalid_1's auc: 0.989582\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[100000]\ttraining's auc: 0.997024\tvalid_1's auc: 0.989582\n",
      "This AUC: 0.9895815070957741\n",
      "Training until validation scores don't improve for 4000 rounds.\n",
      "[5000]\ttraining's auc: 0.946963\tvalid_1's auc: 0.971207\n",
      "[10000]\ttraining's auc: 0.958904\tvalid_1's auc: 0.976338\n",
      "[15000]\ttraining's auc: 0.964882\tvalid_1's auc: 0.978484\n",
      "[20000]\ttraining's auc: 0.969238\tvalid_1's auc: 0.97992\n",
      "[25000]\ttraining's auc: 0.972946\tvalid_1's auc: 0.981127\n",
      "[30000]\ttraining's auc: 0.976176\tvalid_1's auc: 0.982186\n",
      "[35000]\ttraining's auc: 0.979082\tvalid_1's auc: 0.983149\n",
      "[40000]\ttraining's auc: 0.9817\tvalid_1's auc: 0.984057\n",
      "[45000]\ttraining's auc: 0.984041\tvalid_1's auc: 0.984921\n",
      "[50000]\ttraining's auc: 0.986113\tvalid_1's auc: 0.985709\n",
      "[55000]\ttraining's auc: 0.987959\tvalid_1's auc: 0.986451\n",
      "[60000]\ttraining's auc: 0.989578\tvalid_1's auc: 0.987142\n",
      "[65000]\ttraining's auc: 0.991002\tvalid_1's auc: 0.987777\n",
      "[70000]\ttraining's auc: 0.992251\tvalid_1's auc: 0.988356\n",
      "[75000]\ttraining's auc: 0.993337\tvalid_1's auc: 0.988911\n",
      "[80000]\ttraining's auc: 0.994292\tvalid_1's auc: 0.989413\n",
      "[85000]\ttraining's auc: 0.995126\tvalid_1's auc: 0.989889\n",
      "[90000]\ttraining's auc: 0.995847\tvalid_1's auc: 0.990351\n",
      "[95000]\ttraining's auc: 0.996465\tvalid_1's auc: 0.990772\n",
      "[100000]\ttraining's auc: 0.997002\tvalid_1's auc: 0.991158\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[100000]\ttraining's auc: 0.997002\tvalid_1's auc: 0.991158\n",
      "This AUC: 0.9911584549452088\n",
      "Training until validation scores don't improve for 4000 rounds.\n",
      "[5000]\ttraining's auc: 0.947256\tvalid_1's auc: 0.970155\n",
      "[10000]\ttraining's auc: 0.959172\tvalid_1's auc: 0.9755\n",
      "[15000]\ttraining's auc: 0.965158\tvalid_1's auc: 0.977772\n",
      "[20000]\ttraining's auc: 0.969514\tvalid_1's auc: 0.979204\n",
      "[25000]\ttraining's auc: 0.973168\tvalid_1's auc: 0.98041\n",
      "[30000]\ttraining's auc: 0.976383\tvalid_1's auc: 0.981476\n",
      "[35000]\ttraining's auc: 0.979243\tvalid_1's auc: 0.982449\n",
      "[40000]\ttraining's auc: 0.98184\tvalid_1's auc: 0.983365\n",
      "[45000]\ttraining's auc: 0.98414\tvalid_1's auc: 0.984203\n",
      "[50000]\ttraining's auc: 0.986198\tvalid_1's auc: 0.985007\n",
      "[55000]\ttraining's auc: 0.988025\tvalid_1's auc: 0.985743\n",
      "[60000]\ttraining's auc: 0.989635\tvalid_1's auc: 0.986418\n",
      "[65000]\ttraining's auc: 0.991041\tvalid_1's auc: 0.987085\n",
      "[70000]\ttraining's auc: 0.99228\tvalid_1's auc: 0.987673\n",
      "[75000]\ttraining's auc: 0.993365\tvalid_1's auc: 0.988249\n",
      "[80000]\ttraining's auc: 0.994305\tvalid_1's auc: 0.988773\n",
      "[85000]\ttraining's auc: 0.995134\tvalid_1's auc: 0.989261\n",
      "[90000]\ttraining's auc: 0.995839\tvalid_1's auc: 0.989728\n",
      "[95000]\ttraining's auc: 0.996448\tvalid_1's auc: 0.990164\n",
      "[100000]\ttraining's auc: 0.996981\tvalid_1's auc: 0.990582\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[100000]\ttraining's auc: 0.996981\tvalid_1's auc: 0.990582\n",
      "This AUC: 0.9905815602484467\n"
     ]
    }
   ],
   "source": [
    "adasyn_trained_models = get_preds(X_resampled, y_resampled, params)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "save_model(adasyn_trained_models, 'adasyn_aug_lgbm_models_cv10.m')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Borderling SMOTE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "from imblearn.over_sampling import BorderlineSMOTE\n",
    "bsmote = BorderlineSMOTE(random_state=0)\n",
    "X_resampled, y_resampled = bsmote.fit_resample(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training until validation scores don't improve for 4000 rounds.\n",
      "[5000]\ttraining's auc: 0.978414\tvalid_1's auc: 0.82352\n",
      "[10000]\ttraining's auc: 0.985241\tvalid_1's auc: 0.842706\n",
      "[15000]\ttraining's auc: 0.988802\tvalid_1's auc: 0.853939\n",
      "[20000]\ttraining's auc: 0.991253\tvalid_1's auc: 0.861612\n",
      "[25000]\ttraining's auc: 0.993141\tvalid_1's auc: 0.867287\n",
      "[30000]\ttraining's auc: 0.994605\tvalid_1's auc: 0.871503\n",
      "[35000]\ttraining's auc: 0.995776\tvalid_1's auc: 0.875316\n",
      "[40000]\ttraining's auc: 0.996701\tvalid_1's auc: 0.87851\n",
      "[45000]\ttraining's auc: 0.99744\tvalid_1's auc: 0.881267\n",
      "[50000]\ttraining's auc: 0.998021\tvalid_1's auc: 0.883741\n",
      "[55000]\ttraining's auc: 0.99848\tvalid_1's auc: 0.885987\n",
      "[60000]\ttraining's auc: 0.998836\tvalid_1's auc: 0.887958\n",
      "[65000]\ttraining's auc: 0.999114\tvalid_1's auc: 0.889732\n",
      "[70000]\ttraining's auc: 0.999329\tvalid_1's auc: 0.891276\n",
      "[75000]\ttraining's auc: 0.999495\tvalid_1's auc: 0.892637\n",
      "[80000]\ttraining's auc: 0.999623\tvalid_1's auc: 0.893934\n",
      "[85000]\ttraining's auc: 0.999721\tvalid_1's auc: 0.895197\n",
      "[90000]\ttraining's auc: 0.999795\tvalid_1's auc: 0.896355\n",
      "[95000]\ttraining's auc: 0.999851\tvalid_1's auc: 0.897392\n",
      "[100000]\ttraining's auc: 0.999893\tvalid_1's auc: 0.898357\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[100000]\ttraining's auc: 0.999893\tvalid_1's auc: 0.898357\n",
      "This AUC: 0.8983571259578611\n",
      "Training until validation scores don't improve for 4000 rounds.\n",
      "[5000]\ttraining's auc: 0.951605\tvalid_1's auc: 0.976986\n",
      "[10000]\ttraining's auc: 0.963186\tvalid_1's auc: 0.98229\n",
      "[15000]\ttraining's auc: 0.968986\tvalid_1's auc: 0.984817\n",
      "[20000]\ttraining's auc: 0.973106\tvalid_1's auc: 0.98652\n",
      "[25000]\ttraining's auc: 0.976488\tvalid_1's auc: 0.98792\n",
      "[30000]\ttraining's auc: 0.979363\tvalid_1's auc: 0.98911\n",
      "[35000]\ttraining's auc: 0.981884\tvalid_1's auc: 0.990142\n",
      "[40000]\ttraining's auc: 0.9841\tvalid_1's auc: 0.991039\n",
      "[45000]\ttraining's auc: 0.986036\tvalid_1's auc: 0.991817\n",
      "[50000]\ttraining's auc: 0.987743\tvalid_1's auc: 0.992508\n",
      "[55000]\ttraining's auc: 0.98924\tvalid_1's auc: 0.993113\n",
      "[60000]\ttraining's auc: 0.990568\tvalid_1's auc: 0.993632\n",
      "[65000]\ttraining's auc: 0.991735\tvalid_1's auc: 0.9941\n",
      "[70000]\ttraining's auc: 0.992775\tvalid_1's auc: 0.99451\n",
      "[75000]\ttraining's auc: 0.993693\tvalid_1's auc: 0.994877\n",
      "[80000]\ttraining's auc: 0.994513\tvalid_1's auc: 0.995214\n",
      "[85000]\ttraining's auc: 0.995231\tvalid_1's auc: 0.995504\n",
      "[90000]\ttraining's auc: 0.995865\tvalid_1's auc: 0.995765\n",
      "[95000]\ttraining's auc: 0.996431\tvalid_1's auc: 0.996\n",
      "[100000]\ttraining's auc: 0.996921\tvalid_1's auc: 0.99621\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[100000]\ttraining's auc: 0.996921\tvalid_1's auc: 0.99621\n",
      "This AUC: 0.9962101650261468\n",
      "Training until validation scores don't improve for 4000 rounds.\n",
      "[5000]\ttraining's auc: 0.952016\tvalid_1's auc: 0.97582\n",
      "[10000]\ttraining's auc: 0.963564\tvalid_1's auc: 0.981115\n",
      "[15000]\ttraining's auc: 0.969299\tvalid_1's auc: 0.983672\n",
      "[20000]\ttraining's auc: 0.973388\tvalid_1's auc: 0.985381\n",
      "[25000]\ttraining's auc: 0.976717\tvalid_1's auc: 0.986798\n",
      "[30000]\ttraining's auc: 0.979583\tvalid_1's auc: 0.988048\n",
      "[35000]\ttraining's auc: 0.982073\tvalid_1's auc: 0.989114\n",
      "[40000]\ttraining's auc: 0.98425\tvalid_1's auc: 0.990056\n",
      "[45000]\ttraining's auc: 0.986169\tvalid_1's auc: 0.990883\n",
      "[50000]\ttraining's auc: 0.98787\tvalid_1's auc: 0.991616\n",
      "[55000]\ttraining's auc: 0.989365\tvalid_1's auc: 0.992251\n",
      "[60000]\ttraining's auc: 0.990685\tvalid_1's auc: 0.992822\n",
      "[65000]\ttraining's auc: 0.991847\tvalid_1's auc: 0.993318\n",
      "[70000]\ttraining's auc: 0.992868\tvalid_1's auc: 0.993766\n",
      "[75000]\ttraining's auc: 0.993782\tvalid_1's auc: 0.994167\n",
      "[80000]\ttraining's auc: 0.994583\tvalid_1's auc: 0.994533\n",
      "[85000]\ttraining's auc: 0.995295\tvalid_1's auc: 0.994848\n",
      "[90000]\ttraining's auc: 0.995925\tvalid_1's auc: 0.995116\n",
      "[95000]\ttraining's auc: 0.996477\tvalid_1's auc: 0.99539\n",
      "[100000]\ttraining's auc: 0.996965\tvalid_1's auc: 0.995637\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[100000]\ttraining's auc: 0.996965\tvalid_1's auc: 0.995637\n",
      "This AUC: 0.9956371521946755\n",
      "Training until validation scores don't improve for 4000 rounds.\n",
      "[5000]\ttraining's auc: 0.951479\tvalid_1's auc: 0.977433\n",
      "[10000]\ttraining's auc: 0.963099\tvalid_1's auc: 0.982797\n",
      "[15000]\ttraining's auc: 0.968894\tvalid_1's auc: 0.985248\n",
      "[20000]\ttraining's auc: 0.973043\tvalid_1's auc: 0.986895\n",
      "[25000]\ttraining's auc: 0.976425\tvalid_1's auc: 0.98825\n",
      "[30000]\ttraining's auc: 0.97931\tvalid_1's auc: 0.989412\n",
      "[35000]\ttraining's auc: 0.98183\tvalid_1's auc: 0.990431\n",
      "[40000]\ttraining's auc: 0.984045\tvalid_1's auc: 0.991299\n",
      "[45000]\ttraining's auc: 0.985994\tvalid_1's auc: 0.992056\n",
      "[50000]\ttraining's auc: 0.987708\tvalid_1's auc: 0.992722\n",
      "[55000]\ttraining's auc: 0.989206\tvalid_1's auc: 0.993297\n",
      "[60000]\ttraining's auc: 0.990537\tvalid_1's auc: 0.99381\n",
      "[65000]\ttraining's auc: 0.991705\tvalid_1's auc: 0.994259\n",
      "[70000]\ttraining's auc: 0.992739\tvalid_1's auc: 0.99467\n",
      "[75000]\ttraining's auc: 0.993661\tvalid_1's auc: 0.995035\n",
      "[80000]\ttraining's auc: 0.994473\tvalid_1's auc: 0.995355\n",
      "[85000]\ttraining's auc: 0.995196\tvalid_1's auc: 0.995648\n",
      "[90000]\ttraining's auc: 0.995832\tvalid_1's auc: 0.995911\n",
      "[95000]\ttraining's auc: 0.996396\tvalid_1's auc: 0.996147\n",
      "[100000]\ttraining's auc: 0.996895\tvalid_1's auc: 0.996348\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[100000]\ttraining's auc: 0.996895\tvalid_1's auc: 0.996348\n",
      "This AUC: 0.9963481209528733\n",
      "Training until validation scores don't improve for 4000 rounds.\n",
      "[5000]\ttraining's auc: 0.951874\tvalid_1's auc: 0.976581\n",
      "[10000]\ttraining's auc: 0.963405\tvalid_1's auc: 0.981854\n",
      "[15000]\ttraining's auc: 0.969159\tvalid_1's auc: 0.984379\n",
      "[20000]\ttraining's auc: 0.973259\tvalid_1's auc: 0.986137\n",
      "[25000]\ttraining's auc: 0.976614\tvalid_1's auc: 0.987561\n",
      "[30000]\ttraining's auc: 0.979456\tvalid_1's auc: 0.988801\n",
      "[35000]\ttraining's auc: 0.981964\tvalid_1's auc: 0.989869\n",
      "[40000]\ttraining's auc: 0.984155\tvalid_1's auc: 0.990796\n",
      "[45000]\ttraining's auc: 0.986083\tvalid_1's auc: 0.991607\n",
      "[50000]\ttraining's auc: 0.987789\tvalid_1's auc: 0.992318\n",
      "[55000]\ttraining's auc: 0.989273\tvalid_1's auc: 0.992951\n",
      "[60000]\ttraining's auc: 0.990602\tvalid_1's auc: 0.993498\n",
      "[65000]\ttraining's auc: 0.991765\tvalid_1's auc: 0.993995\n",
      "[70000]\ttraining's auc: 0.992795\tvalid_1's auc: 0.994434\n",
      "[75000]\ttraining's auc: 0.993704\tvalid_1's auc: 0.994814\n",
      "[80000]\ttraining's auc: 0.994516\tvalid_1's auc: 0.995159\n",
      "[85000]\ttraining's auc: 0.995237\tvalid_1's auc: 0.995464\n",
      "[90000]\ttraining's auc: 0.995865\tvalid_1's auc: 0.995732\n",
      "[95000]\ttraining's auc: 0.99642\tvalid_1's auc: 0.995985\n",
      "[100000]\ttraining's auc: 0.996922\tvalid_1's auc: 0.996206\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[100000]\ttraining's auc: 0.996922\tvalid_1's auc: 0.996206\n",
      "This AUC: 0.9962057938741212\n"
     ]
    }
   ],
   "source": [
    "bsmote_trained_models = get_preds(X_resampled, y_resampled, params)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "save_model(bsmote_trained_models, 'bsmote_aug_lgbm_models_cv10.m')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shutdown scheduled for Sat 2019-03-23 04:48:54 UTC, use 'shutdown -c' to cancel.\r\n"
     ]
    }
   ],
   "source": [
    "!shutdown"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}