h2o-rf-gbm-stacked.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Loan default prediction - binary classification\n",
    "\n",
    "* Libraries, data preparation\n",
    "* Random Forest - Hyperparameters, Grid-search\n",
    "* GBM - Hyperparameters, Grid-search\n",
    "* Stacked Ensembles\n",
    "* Comparing Models\n",
    "* Conclusion\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####For this project I specifically focus less on feature engineering and use a small dataset. The main purpose is to build supervised machine learning models with h2o and experiment with hyperparameters."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import datetime\n",
    "import datetime\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####Load the data and create train and test dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = pd.read_csv(\"loandefault.csv\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####I will create a month variable from AP005 (date) column and I will drop AP005 column\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "data['AP005']= pd.to_datetime(data['AP005']) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "data['month'] = data['AP005'].dt.month"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####I will drop the following variables"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = data.drop(columns=['id', 'Unnamed: 0', 'Unnamed: 0.1', 'AP005', 'TD025', 'TD026', 'TD027', 'TD028', 'TD029',\n",
    "                          'TD044', 'TD048', 'TD051', 'TD054', 'TD055', 'TD061', 'TD062', 'CR012']) #'TD048', 'TD055', 'TD062', 'TD044', 'TD051', 'TD061', 'TD054',"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####Binning the age column"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "data['AP001'] = pd.cut(x=data['AP001'],  bins=[19, 24, 29, 34, 39, 44, 49, 54, 59])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(48000, 73)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "train, test = train_test_split(\n",
    "     data, test_size=0.40, random_state=8)\n",
    "train.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####Assign the index, columns, source and count variable in each source"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "CD    36\n",
       "TD    12\n",
       "AP     8\n",
       "CR     7\n",
       "PA     6\n",
       "MB     2\n",
       "lo     1\n",
       "mo     1\n",
       "Name: source, dtype: int64"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "var = pd.DataFrame(train.dtypes)\n",
    "var = pd.DataFrame(train.dtypes).reset_index()\n",
    "var.columns = ['varname','dtype'] \n",
    "var['source'] = var['varname'].str[:2]\n",
    "#var.head()\n",
    "var['source'].value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####Create the list of variables"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['AP001', 'AP002', 'AP003', 'AP004', 'AP006', 'AP007', 'AP008', 'AP009']"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "MB_list = list(var[var['source']=='MB']['varname'])\n",
    "AP_list = list(var[var['source']=='AP']['varname'])\n",
    "TD_list = list(var[var['source']=='TD']['varname'])\n",
    "CR_list = list(var[var['source']=='CR']['varname'])\n",
    "PA_list = list(var[var['source']=='PA']['varname'])\n",
    "CD_list = list(var[var['source']=='CD']['varname'])\n",
    "AP_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    38661\n",
       "1     9339\n",
       "Name: loan_default, dtype: int64"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train['loan_default'].value_counts(dropna=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###Initialize H2O"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Checking whether there is an H2O instance running at http://localhost:54321 . connected.\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td>H2O cluster uptime:</td>\n",
       "<td>4 days 23 hours 59 mins</td></tr>\n",
       "<tr><td>H2O cluster timezone:</td>\n",
       "<td>America/New_York</td></tr>\n",
       "<tr><td>H2O data parsing timezone:</td>\n",
       "<td>UTC</td></tr>\n",
       "<tr><td>H2O cluster version:</td>\n",
       "<td>3.26.0.10</td></tr>\n",
       "<tr><td>H2O cluster version age:</td>\n",
       "<td>22 days </td></tr>\n",
       "<tr><td>H2O cluster name:</td>\n",
       "<td>H2O_from_python_natab_bcm2mk</td></tr>\n",
       "<tr><td>H2O cluster total nodes:</td>\n",
       "<td>1</td></tr>\n",
       "<tr><td>H2O cluster free memory:</td>\n",
       "<td>2.528 Gb</td></tr>\n",
       "<tr><td>H2O cluster total cores:</td>\n",
       "<td>8</td></tr>\n",
       "<tr><td>H2O cluster allowed cores:</td>\n",
       "<td>8</td></tr>\n",
       "<tr><td>H2O cluster status:</td>\n",
       "<td>locked, healthy</td></tr>\n",
       "<tr><td>H2O connection url:</td>\n",
       "<td>http://localhost:54321</td></tr>\n",
       "<tr><td>H2O connection proxy:</td>\n",
       "<td>{'http': None, 'https': None}</td></tr>\n",
       "<tr><td>H2O internal security:</td>\n",
       "<td>False</td></tr>\n",
       "<tr><td>H2O API Extensions:</td>\n",
       "<td>Amazon S3, Algos, AutoML, Core V3, TargetEncoder, Core V4</td></tr>\n",
       "<tr><td>Python version:</td>\n",
       "<td>3.7.3 final</td></tr></table></div>"
      ],
      "text/plain": [
       "--------------------------  ---------------------------------------------------------\n",
       "H2O cluster uptime:         4 days 23 hours 59 mins\n",
       "H2O cluster timezone:       America/New_York\n",
       "H2O data parsing timezone:  UTC\n",
       "H2O cluster version:        3.26.0.10\n",
       "H2O cluster version age:    22 days\n",
       "H2O cluster name:           H2O_from_python_natab_bcm2mk\n",
       "H2O cluster total nodes:    1\n",
       "H2O cluster free memory:    2.528 Gb\n",
       "H2O cluster total cores:    8\n",
       "H2O cluster allowed cores:  8\n",
       "H2O cluster status:         locked, healthy\n",
       "H2O connection url:         http://localhost:54321\n",
       "H2O connection proxy:       {'http': None, 'https': None}\n",
       "H2O internal security:      False\n",
       "H2O API Extensions:         Amazon S3, Algos, AutoML, Core V3, TargetEncoder, Core V4\n",
       "Python version:             3.7.3 final\n",
       "--------------------------  ---------------------------------------------------------"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import h2o\n",
    "h2o.init()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "from h2o.estimators.gbm import H2OGradientBoostingEstimator\n",
    "from h2o.estimators.random_forest import H2ORandomForestEstimator"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####I will create train and test sample and convert it into h2o dataframe"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n"
     ]
    }
   ],
   "source": [
    "train_smpl = train.sample(frac=0.1, random_state=1)\n",
    "test_smpl = test.sample(frac=0.1, random_state=1)\n",
    "train_hex = h2o.H2OFrame(train_smpl)\n",
    "test_hex = h2o.H2OFrame(test_smpl)\n",
    "#train_hex.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####Identify response and predictor variables"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "target='loan_default'\n",
    "predictors = CR_list + TD_list + AP_list + MB_list + CR_list + PA_list \n",
    "#predictors = newlist"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####I am trying to predit a binary classification problem, so I need response variable to be a factor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_hex['loan_default'] = train_hex['loan_default'].asfactor()\n",
    "test_hex['loan_default'] = test_hex['loan_default'].asfactor()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####RANDOM FOREST - is an ensemble learning method, composed of multiple decision trees. By averaging out the impact of several decision trees, random forests tend to improve prediction."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####I will train base random forest model  WITH default settings FIRST, the goal of our initial model is to provide a baseline AUC that I will try later to beat with Hyperparameter tunning. The default number of trees in an H2O Random Forest is 50, usually increasing the trees in random forest increase the performance."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "nfolds = 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "drf Model Build progress: |███████████████████████████████████████████████| 100%\n"
     ]
    }
   ],
   "source": [
    "rf_v1 = H2ORandomForestEstimator(\n",
    "        model_id = 'rf_v1',\n",
    "        nfolds=nfolds,\n",
    "        seed=8)\n",
    "rf_v1.train(predictors,target,training_frame=train_hex)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 418,
   "metadata": {},
   "outputs": [],
   "source": [
    "#rf_v1.confusion_matrix()\n",
    "#rf_v1.auc() "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "rf_v1.varimp_plot(num_of_features=25)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 479,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "rf_v1.test = rf_v1.model_performance(test_hex)\n",
    "rf_v1.test.plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####To get the best performance l want to tune at least some of these hyperparameters to maximize the predictive performance of  model. I would love to do full grid of hyperparameters, but I found that this could take a very substantial time to run in practice. I decided to focus on on most important hyperparameters: n_estimators, max_features, and max_depth.\n",
    "\n",
    "####We can try out hyperparameter sets by hand - (manual search) - individually train a number of models with different combinations of values, and then to compare the model performance to choose the best model or we can use the Grid search - it builds models for every combination of hyperparameter values that we specify.\n",
    "\n",
    "\n",
    "####adjusting the following set of hyperparameters:\n",
    "\n",
    "- ntrees = number of trees in the foreset\n",
    "- max_depth = max number of levels in each decision tree\n",
    "- min_rows = the minimum number of observations for a leaf \n",
    "- mtries = the columns to randomly select at each level.\n",
    "\n",
    "I will first create the Random Hyperparameter Grid\n",
    "On each iteration, the algorithm will choose a difference combination of the features."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###GRID SEARCH"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 324,
   "metadata": {},
   "outputs": [],
   "source": [
    "import h2o.grid\n",
    "\n",
    "g = h2o.grid.H2OGridSearch(\n",
    "  h2o.estimators.H2ORandomForestEstimator(\n",
    "    nfolds=5\n",
    "    ),\n",
    "  hyper_params={\n",
    "    \"ntrees\": [200, 800],\n",
    "    \"max_depth\": [40, 60],\n",
    "    \"min_rows\": [80, 100, 200],\n",
    "      \"mtries\": [2, 8, 10]\n",
    "    }\n",
    "  )\n",
    "#g.train(target, predictors, train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "drf Grid Build progress: |████████████████████████████████████████████████| 100%\n"
     ]
    }
   ],
   "source": [
    "g.train(predictors, target, train_hex)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####by default, display the grid search results sorted by increasing logloss (we can change this parameter and sort it by auc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "     max_depth min_rows mtries ntrees  \\\n",
      "0           40     80.0     10    800   \n",
      "1           60     80.0     10    200   \n",
      "2           40     80.0     10    200   \n",
      "3           60     80.0     10    800   \n",
      "4           60    100.0      8    200   \n",
      "5           40    100.0     10    800   \n",
      "6           40     80.0      8    200   \n",
      "7           40     80.0      8    800   \n",
      "8           60     80.0      8    800   \n",
      "9           40    100.0      8    800   \n",
      "10          40    100.0     10    200   \n",
      "11          60    100.0     10    800   \n",
      "12          60     80.0      8    200   \n",
      "13          60    100.0     10    200   \n",
      "14          60    100.0      8    800   \n",
      "15          60    200.0     10    800   \n",
      "16          40    100.0      8    200   \n",
      "17          60    200.0     10    200   \n",
      "18          40    200.0     10    200   \n",
      "19          40    200.0     10    800   \n",
      "20          60    200.0      8    200   \n",
      "21          40    200.0      8    200   \n",
      "22          60    200.0      8    800   \n",
      "23          40    200.0      8    800   \n",
      "24          60     80.0      2    800   \n",
      "25          60     80.0      2    200   \n",
      "26          40     80.0      2    800   \n",
      "27          40     80.0      2    200   \n",
      "28          40    100.0      2    800   \n",
      "29          40    100.0      2    200   \n",
      "30          60    100.0      2    800   \n",
      "31          60    100.0      2    200   \n",
      "32          40    200.0      2    800   \n",
      "33          40    200.0      2    200   \n",
      "34          60    200.0      2    800   \n",
      "35          60    200.0      2    200   \n",
      "\n",
      "                                                            model_ids  \\\n",
      "0   Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_31   \n",
      "1   Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_14   \n",
      "2   Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_13   \n",
      "3   Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_32   \n",
      "4   Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_10   \n",
      "5   Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_33   \n",
      "6    Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_7   \n",
      "7   Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_25   \n",
      "8   Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_26   \n",
      "9   Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_27   \n",
      "10  Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_15   \n",
      "11  Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_34   \n",
      "12   Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_8   \n",
      "13  Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_16   \n",
      "14  Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_28   \n",
      "15  Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_36   \n",
      "16   Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_9   \n",
      "17  Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_18   \n",
      "18  Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_17   \n",
      "19  Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_35   \n",
      "20  Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_12   \n",
      "21  Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_11   \n",
      "22  Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_30   \n",
      "23  Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_29   \n",
      "24  Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_20   \n",
      "25   Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_2   \n",
      "26  Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_19   \n",
      "27   Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_1   \n",
      "28  Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_21   \n",
      "29   Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_3   \n",
      "30  Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_22   \n",
      "31   Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_4   \n",
      "32  Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_23   \n",
      "33   Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_5   \n",
      "34  Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_24   \n",
      "35   Grid_DRF_py_1_sid_aa33_model_python_1573941761911_203736_model_6   \n",
      "\n",
      "                logloss  \n",
      "0   0.49029129124238163  \n",
      "1     0.491244291414509  \n",
      "2     0.491325758441785  \n",
      "3    0.4914339008395886  \n",
      "4    0.4915413829624825  \n",
      "5    0.4916901261254847  \n",
      "6    0.4918717345471083  \n",
      "7    0.4920097004468391  \n",
      "8    0.4920415563529053  \n",
      "9   0.49221308807907765  \n",
      "10  0.49227164738877566  \n",
      "11   0.4923767525273119  \n",
      "12   0.4923924812478064  \n",
      "13   0.4925426647464404  \n",
      "14   0.4928778356189695  \n",
      "15  0.49438450576059534  \n",
      "16   0.4944642612869156  \n",
      "17  0.49483140133688636  \n",
      "18  0.49497103071552945  \n",
      "19   0.4949841023225295  \n",
      "20   0.4951051107062385  \n",
      "21  0.49516549462976733  \n",
      "22  0.49545838946197507  \n",
      "23   0.4960540599575154  \n",
      "24   0.4975616421387109  \n",
      "25  0.49773911060103776  \n",
      "26  0.49780050994871144  \n",
      "27  0.49798837851114985  \n",
      "28   0.4985035785582407  \n",
      "29   0.4987329001155938  \n",
      "30  0.49897820780095437  \n",
      "31  0.49928844343308393  \n",
      "32    0.501759180844455  \n",
      "33   0.5018357308439517  \n",
      "34   0.5021699635435108  \n",
      "35   0.5025184799544471  \n"
     ]
    },
    {
     "data": {
      "text/plain": []
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "g "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####It has ordered from best to worst, we want logloss to be low. My results are: ‘max_depth’ = 40; ‘min_rows’ = ‘80’; ‘mtries’: 10 and 'ntrees' : 800, Now I WILL plug these back into the model to see if it improved  performance."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 480,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "drf Model Build progress: |███████████████████████████████████████████████| 100%\n"
     ]
    }
   ],
   "source": [
    "rf_tunned = H2ORandomForestEstimator(\n",
    "        model_id = 'rf_tunned',\n",
    "        ntrees = 800,\n",
    "        min_rows=80,\n",
    "        max_depth=40,\n",
    "        nfolds = nfolds,\n",
    "        fold_assignment = \"Modulo\",\n",
    "        keep_cross_validation_predictions = True,\n",
    "        seed=8)\n",
    "rf_tunned.train(predictors,target,training_frame=train_hex)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 422,
   "metadata": {},
   "outputs": [],
   "source": [
    "#rf_tunned.auc()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 481,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "rf_tunned.test = rf_tunned.model_performance(test_hex)\n",
    "rf_tunned.test.plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####The auc for our rf base model on train set was 63.55, after the tunning it improved 0.6565. the Grater the AUC the better is the performance of our model. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####GBM h2o’s GBM sequentially builds regression trees on all the features of the dataset in a fully distributed way - each tree is built in parallel. First I will build a base model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 498,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "gbm Model Build progress: |███████████████████████████████████████████████| 100%\n"
     ]
    }
   ],
   "source": [
    "gbm_v1 = H2OGradientBoostingEstimator(\n",
    "        model_id = 'gbm_v1',\n",
    "        nfolds = nfolds,\n",
    "        fold_assignment = \"Modulo\",\n",
    "        keep_cross_validation_predictions = True,\n",
    "        seed=8)\n",
    "gbm_v1.train(predictors,target,training_frame=train_hex)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 517,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1400x1000 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "gbm_v1.varimp_plot(num_of_features=20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 500,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "gbm_v1.test = gbm_v1.model_performance(test_hex)\n",
    "gbm_v1.test.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "from h2o.estimators.gbm import H2OGradientBoostingEstimator\n",
    "from h2o.grid.grid_search import H2OGridSearch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 425,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Hyperparameters\n",
    "gbm_params1 = {'learn_rate': [0.001, 0.01],\n",
    "               'ntrees' : [10, 20, 50, 100, 500],\n",
    "                'max_depth': [3, 5, 9],\n",
    "                #'sample_rate': [0.4, 0.8, 1.0],\n",
    "                #'col_sample_rate': [0.2, 0.5, 1.0],\n",
    "               'min_rows' : [50, 100, 200]\n",
    "                }"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 426,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "gbm Grid Build progress: |████████████████████████████████████████████████| 100%\n",
      "       col_sample_rate learn_rate max_depth sample_rate            model_ids  \\\n",
      "0                  0.5       0.01         3         0.8   gbm_grid1_model_38   \n",
      "1                  0.2       0.01         3         0.8   gbm_grid1_model_37   \n",
      "2                  0.2       0.01         3         1.0   gbm_grid1_model_55   \n",
      "3                  0.5       0.01         3         1.0   gbm_grid1_model_56   \n",
      "4                  1.0        0.1         3         0.8    gbm_grid1_model_6   \n",
      "5                  1.0       0.01         3         0.8   gbm_grid1_model_39   \n",
      "6                  0.2        0.1         3         1.0   gbm_grid1_model_22   \n",
      "7                  0.5       0.01         5         0.8    gbm_grid1_model_8   \n",
      "8                  0.5       0.01         5         0.8   gbm_grid1_model_44   \n",
      "9                  0.2       0.01         5         0.8   gbm_grid1_model_43   \n",
      "10                 0.2        0.1         3         0.8    gbm_grid1_model_4   \n",
      "11                 1.0       0.01         3         1.0   gbm_grid1_model_57   \n",
      "12                 1.0       0.01         3         1.0  gbm_grid1_model_146   \n",
      "13                 0.2       0.01         5         1.0   gbm_grid1_model_25   \n",
      "14                 0.2       0.01         5         0.8    gbm_grid1_model_7   \n",
      "15                 0.5       0.01         5         1.0   gbm_grid1_model_62   \n",
      "16                 0.2       0.01         5         1.0   gbm_grid1_model_61   \n",
      "17                 0.2       0.01         3         1.0   gbm_grid1_model_19   \n",
      "18                 1.0       0.01         3         1.0   gbm_grid1_model_21   \n",
      "19                 1.0       0.01         3         1.0  gbm_grid1_model_152   \n",
      "20                 0.2       0.01         3         0.8    gbm_grid1_model_1   \n",
      "21                 0.5       0.01         9         0.8   gbm_grid1_model_14   \n",
      "22                 1.0       0.01         5         0.8   gbm_grid1_model_45   \n",
      "23                 1.0       0.01         3         0.8    gbm_grid1_model_3   \n",
      "24                 1.0       0.01         5         0.8    gbm_grid1_model_9   \n",
      "25                 1.0       0.01         3         1.0  gbm_grid1_model_158   \n",
      "26                 0.5       0.01         3         0.8    gbm_grid1_model_2   \n",
      "27                 0.5       0.01         3         1.0   gbm_grid1_model_20   \n",
      "28                 0.2       0.01         9         1.0   gbm_grid1_model_31   \n",
      "29                 0.5       0.01         5         1.0   gbm_grid1_model_26   \n",
      "..  ..             ...        ...       ...         ...                  ...   \n",
      "132                1.0        0.1         9         1.0   gbm_grid1_model_72   \n",
      "133                1.0      0.001         9         1.0  gbm_grid1_model_125   \n",
      "134                1.0       0.01         3         1.0   gbm_grid1_model_74   \n",
      "135                1.0        0.1         9         0.8   gbm_grid1_model_54   \n",
      "136                1.0      0.001         3         1.0  gbm_grid1_model_127   \n",
      "137                1.0      0.001         5         1.0  gbm_grid1_model_123   \n",
      "138                1.0        0.1         9         1.0   gbm_grid1_model_36   \n",
      "139                1.0      0.001         3         1.0  gbm_grid1_model_133   \n",
      "140                0.5        0.1         9         1.0   gbm_grid1_model_71   \n",
      "141                1.0      0.001         9         1.0   gbm_grid1_model_95   \n",
      "142                1.0      0.001         9         1.0   gbm_grid1_model_77   \n",
      "143                1.0       0.01         3         1.0   gbm_grid1_model_80   \n",
      "144                0.2        0.1         5         0.8   gbm_grid1_model_46   \n",
      "145                1.0      0.001         3         1.0  gbm_grid1_model_115   \n",
      "146                1.0      0.001         3         1.0   gbm_grid1_model_97   \n",
      "147                1.0      0.001         3         1.0   gbm_grid1_model_79   \n",
      "148                1.0      0.001         3         1.0  gbm_grid1_model_109   \n",
      "149                1.0      0.001         3         1.0   gbm_grid1_model_91   \n",
      "150                1.0      0.001         3         1.0   gbm_grid1_model_73   \n",
      "151                1.0      0.001         9         1.0  gbm_grid1_model_107   \n",
      "152                1.0      0.001         5         1.0  gbm_grid1_model_105   \n",
      "153                1.0      0.001         5         1.0   gbm_grid1_model_87   \n",
      "154                1.0      0.001         9         1.0   gbm_grid1_model_89   \n",
      "155                1.0      0.001         3         1.0   gbm_grid1_model_85   \n",
      "156                1.0      0.001         3         1.0  gbm_grid1_model_103   \n",
      "157                0.5        0.1         5         1.0   gbm_grid1_model_65   \n",
      "158                1.0        0.1         5         0.8   gbm_grid1_model_48   \n",
      "159                0.2        0.1         5         1.0   gbm_grid1_model_64   \n",
      "160                0.5        0.1         5         0.8   gbm_grid1_model_47   \n",
      "161                1.0        0.1         5         1.0   gbm_grid1_model_66   \n",
      "\n",
      "                    auc  \n",
      "0    0.6735093462039776  \n",
      "1    0.6731300497972357  \n",
      "2    0.6726337800181734  \n",
      "3    0.6692885307450793  \n",
      "4    0.6690602592583952  \n",
      "5    0.6686869534017181  \n",
      "6    0.6682934688500856  \n",
      "7    0.6681222652350723  \n",
      "8    0.6680160117944473  \n",
      "9    0.6678009825763275  \n",
      "10   0.6676474353194004  \n",
      "11    0.666793624289095  \n",
      "12   0.6665908914632139  \n",
      "13   0.6654425976034016  \n",
      "14   0.6653798544737742  \n",
      "15   0.6642492169720481  \n",
      "16   0.6640691851779918  \n",
      "17   0.6636460631681433  \n",
      "18   0.6635517908276481  \n",
      "19   0.6634269351526109  \n",
      "20   0.6633005030170303  \n",
      "21   0.6630898878884319  \n",
      "22   0.6627481112426229  \n",
      "23    0.662637759004585  \n",
      "24   0.6623795347675762  \n",
      "25   0.6622597237662778  \n",
      "26   0.6619936172265519  \n",
      "27    0.661646795907004  \n",
      "28   0.6611833165072444  \n",
      "29   0.6609768001760592  \n",
      "..                  ...  \n",
      "132  0.6210472616565069  \n",
      "133   0.619859240991001  \n",
      "134  0.6188843577909626  \n",
      "135  0.6187563491948386  \n",
      "136  0.6187109471311886  \n",
      "137  0.6183969161909434  \n",
      "138  0.6181651764910637  \n",
      "139  0.6178558749324486  \n",
      "140  0.6171146231849421  \n",
      "141  0.6159959667833458  \n",
      "142  0.6158890827585033  \n",
      "143  0.6149671686327232  \n",
      "144  0.6131990104872461  \n",
      "145  0.6127427828059863  \n",
      "146  0.6127251264479002  \n",
      "147  0.6127251264479002  \n",
      "148  0.6108371573011248  \n",
      "149  0.6093048376529403  \n",
      "150  0.6093048376529403  \n",
      "151  0.6083167121843375  \n",
      "152  0.6083167121843375  \n",
      "153  0.6083167121843375  \n",
      "154  0.6083167121843375  \n",
      "155  0.6081940635540609  \n",
      "156  0.6081940635540609  \n",
      "157   0.607460063525054  \n",
      "158  0.6063669457842608  \n",
      "159  0.6053936390447658  \n",
      "160  0.6009968905892242  \n",
      "161  0.5993485434450455  \n",
      "\n",
      "[162 rows x 7 columns]\n"
     ]
    },
    {
     "data": {
      "text/plain": []
     },
     "execution_count": 426,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Grid search\n",
    "gbm_grid1 = H2OGridSearch(model=H2OGradientBoostingEstimator,\n",
    "                          grid_id='gbm_grid1',\n",
    "                          hyper_params=gbm_params1)\n",
    "gbm_grid1.train(x=predictors, y=target,\n",
    "                training_frame=train_hex,\n",
    "                validation_frame=test_hex,\n",
    "                ntrees = 1000,\n",
    "                seed=8)\n",
    "\n",
    "#grid results, sorted by validation AUC\n",
    "gbm_gridperf1 = gbm_grid1.get_grid(sort_by='auc', decreasing=True)\n",
    "gbm_gridperf1\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####Now i will use the hyperparameters from grid search"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 482,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "gbm Model Build progress: |███████████████████████████████████████████████| 100%\n"
     ]
    }
   ],
   "source": [
    "gbm_tunned = H2OGradientBoostingEstimator(\n",
    "        model_id = 'gbm_tunned',\n",
    "        col_sample_rate = 0.5,\n",
    "        learn_rate = 0.01,\n",
    "        sample_rate = 0.8,\n",
    "        min_rows = 200,\n",
    "        max_depth= 5,\n",
    "        ntrees = 1000,\n",
    "        nfolds = nfolds,\n",
    "        fold_assignment = \"Modulo\",\n",
    "        keep_cross_validation_predictions = True,\n",
    "        seed=8)\n",
    "gbm_tunned.train(predictors,target,training_frame=train_hex)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 501,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "gbm_tunned.test = gbm_tunned.model_performance(test_hex)\n",
    "gbm_tunned.test.plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####After tunning GBM model auc slightly increased from 0.6498 to 0.6569"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####Stacked Ensambles - Ensemble machine learning methods use multiple learning algorithms to obtain better predictive performance than could be obtained from any of the constituent learning algorithms. I will use our best performed (tunned) GBM and RF models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 502,
   "metadata": {},
   "outputs": [],
   "source": [
    "from h2o.estimators.stackedensemble import H2OStackedEnsembleEstimator"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 503,
   "metadata": {},
   "outputs": [],
   "source": [
    "nfolds = 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 504,
   "metadata": {},
   "outputs": [],
   "source": [
    "models = [gbm_tunned.model_id, rf_tunned.model_id]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 505,
   "metadata": {},
   "outputs": [],
   "source": [
    "m_SE = H2OStackedEnsembleEstimator(model_id=\"SE_glm_rf\",\n",
    "                                  base_models = models)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 506,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble Model Build progress: |███████████████████████████████████| 100%\n"
     ]
    }
   ],
   "source": [
    "m_SE.train(predictors, target, train_hex)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####Stacked Ensamble"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 507,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "m_SE.test = m_SE.model_performance(test_hex)\n",
    "m_SE.test.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 508,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 509,
   "metadata": {},
   "outputs": [],
   "source": [
    "all_models = [gbm_tunned, rf_tunned, m_SE]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 510,
   "metadata": {},
   "outputs": [],
   "source": [
    "names = [\"GBM\", \"RF\", \"SE\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####Check performance of these models on test set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 511,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_perf = list(map(lambda x: x.model_performance(test_hex), all_models))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###Logloss for GBM, RF, SE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 512,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GBM    0.466709\n",
       "RF     0.469079\n",
       "SE     0.465213\n",
       "dtype: float64"
      ]
     },
     "execution_count": 512,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.Series(map(lambda p: p.logloss(), test_perf), names)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####Ensemble model gives slighltly better results compared to gbm and rf models individually with lower logloss value"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###AUC for GBM, RF, SE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 513,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GBM    0.656902\n",
       "RF     0.656512\n",
       "SE     0.662925\n",
       "dtype: float64"
      ]
     },
     "execution_count": 513,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.Series(map(lambda p: p.auc(), test_perf), names)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####Again Ensemble model gives slighltly better results compared to both models individually, it has higher AUC  values,  AUC =  0.662925"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 530,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Confusion Matrix (Act/Pred) for max f1 @ threshold = 0.21197437442158915: \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></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>Error</th>\n",
       "      <th>Rate</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>2283.0</td>\n",
       "      <td>1515.0</td>\n",
       "      <td>0.3989</td>\n",
       "      <td>(1515.0/3798.0)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>389.0</td>\n",
       "      <td>613.0</td>\n",
       "      <td>0.3882</td>\n",
       "      <td>(389.0/1002.0)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Total</td>\n",
       "      <td>2672.0</td>\n",
       "      <td>2128.0</td>\n",
       "      <td>0.3967</td>\n",
       "      <td>(1904.0/4800.0)</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               0       1   Error              Rate\n",
       "0      0  2283.0  1515.0  0.3989   (1515.0/3798.0)\n",
       "1      1   389.0   613.0  0.3882    (389.0/1002.0)\n",
       "2  Total  2672.0  2128.0  0.3967   (1904.0/4800.0)"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": []
     },
     "execution_count": 530,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rf_tunned.confusion_matrix()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 531,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Confusion Matrix (Act/Pred) for max f1 @ threshold = 0.23927467285579945: \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></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>Error</th>\n",
       "      <th>Rate</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>2869.0</td>\n",
       "      <td>929.0</td>\n",
       "      <td>0.2446</td>\n",
       "      <td>(929.0/3798.0)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>334.0</td>\n",
       "      <td>668.0</td>\n",
       "      <td>0.3333</td>\n",
       "      <td>(334.0/1002.0)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Total</td>\n",
       "      <td>3203.0</td>\n",
       "      <td>1597.0</td>\n",
       "      <td>0.2631</td>\n",
       "      <td>(1263.0/4800.0)</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               0       1   Error              Rate\n",
       "0      0  2869.0   929.0  0.2446    (929.0/3798.0)\n",
       "1      1   334.0   668.0  0.3333    (334.0/1002.0)\n",
       "2  Total  3203.0  1597.0  0.2631   (1263.0/4800.0)"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": []
     },
     "execution_count": 531,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gbm_tunned.confusion_matrix()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 532,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Confusion Matrix (Act/Pred) for max f1 @ threshold = 0.23138219885026245: \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></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>Error</th>\n",
       "      <th>Rate</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>2926.0</td>\n",
       "      <td>872.0</td>\n",
       "      <td>0.2296</td>\n",
       "      <td>(872.0/3798.0)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>385.0</td>\n",
       "      <td>617.0</td>\n",
       "      <td>0.3842</td>\n",
       "      <td>(385.0/1002.0)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Total</td>\n",
       "      <td>3311.0</td>\n",
       "      <td>1489.0</td>\n",
       "      <td>0.2619</td>\n",
       "      <td>(1257.0/4800.0)</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               0       1   Error              Rate\n",
       "0      0  2926.0   872.0  0.2296    (872.0/3798.0)\n",
       "1      1   385.0   617.0  0.3842    (385.0/1002.0)\n",
       "2  Total  3311.0  1489.0  0.2619   (1257.0/4800.0)"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": []
     },
     "execution_count": 532,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "m_SE.confusion_matrix()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 518,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "drf prediction progress: |████████████████████████████████████████████████| 100%\n"
     ]
    }
   ],
   "source": [
    "predictions = rf_tunned.predict(test_hex)\n",
    "predictions.head()\n",
    "test_scores = test_hex['loan_default'].cbind(predictions).as_data_frame()\n",
    "#test_scores.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 519,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "drf prediction progress: |████████████████████████████████████████████████| 100%\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>count</th>\n",
       "      <th>actual</th>\n",
       "      <th>non_actual</th>\n",
       "      <th>cum_count</th>\n",
       "      <th>cum_actual</th>\n",
       "      <th>cum_non_actual</th>\n",
       "      <th>percent_cum_actual</th>\n",
       "      <th>percent_cum_non_actual</th>\n",
       "      <th>if_random</th>\n",
       "      <th>lift</th>\n",
       "      <th>K_S</th>\n",
       "      <th>gain</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>decile</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>320</td>\n",
       "      <td>80</td>\n",
       "      <td>240</td>\n",
       "      <td>320</td>\n",
       "      <td>80</td>\n",
       "      <td>240</td>\n",
       "      <td>0.13</td>\n",
       "      <td>0.09</td>\n",
       "      <td>61.3</td>\n",
       "      <td>1.31</td>\n",
       "      <td>4.0</td>\n",
       "      <td>25.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>320</td>\n",
       "      <td>89</td>\n",
       "      <td>231</td>\n",
       "      <td>640</td>\n",
       "      <td>169</td>\n",
       "      <td>471</td>\n",
       "      <td>0.28</td>\n",
       "      <td>0.18</td>\n",
       "      <td>122.6</td>\n",
       "      <td>1.38</td>\n",
       "      <td>10.0</td>\n",
       "      <td>26.41</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>320</td>\n",
       "      <td>86</td>\n",
       "      <td>234</td>\n",
       "      <td>960</td>\n",
       "      <td>255</td>\n",
       "      <td>705</td>\n",
       "      <td>0.42</td>\n",
       "      <td>0.27</td>\n",
       "      <td>183.9</td>\n",
       "      <td>1.39</td>\n",
       "      <td>15.0</td>\n",
       "      <td>26.56</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>320</td>\n",
       "      <td>75</td>\n",
       "      <td>245</td>\n",
       "      <td>1280</td>\n",
       "      <td>330</td>\n",
       "      <td>950</td>\n",
       "      <td>0.54</td>\n",
       "      <td>0.37</td>\n",
       "      <td>245.2</td>\n",
       "      <td>1.35</td>\n",
       "      <td>17.0</td>\n",
       "      <td>25.78</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>320</td>\n",
       "      <td>81</td>\n",
       "      <td>239</td>\n",
       "      <td>1600</td>\n",
       "      <td>411</td>\n",
       "      <td>1189</td>\n",
       "      <td>0.67</td>\n",
       "      <td>0.46</td>\n",
       "      <td>306.5</td>\n",
       "      <td>1.34</td>\n",
       "      <td>21.0</td>\n",
       "      <td>25.69</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>320</td>\n",
       "      <td>39</td>\n",
       "      <td>281</td>\n",
       "      <td>1920</td>\n",
       "      <td>450</td>\n",
       "      <td>1470</td>\n",
       "      <td>0.73</td>\n",
       "      <td>0.57</td>\n",
       "      <td>367.8</td>\n",
       "      <td>1.22</td>\n",
       "      <td>16.0</td>\n",
       "      <td>23.44</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>320</td>\n",
       "      <td>30</td>\n",
       "      <td>290</td>\n",
       "      <td>2240</td>\n",
       "      <td>480</td>\n",
       "      <td>1760</td>\n",
       "      <td>0.78</td>\n",
       "      <td>0.68</td>\n",
       "      <td>429.1</td>\n",
       "      <td>1.12</td>\n",
       "      <td>10.0</td>\n",
       "      <td>21.43</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>320</td>\n",
       "      <td>45</td>\n",
       "      <td>275</td>\n",
       "      <td>2560</td>\n",
       "      <td>525</td>\n",
       "      <td>2035</td>\n",
       "      <td>0.86</td>\n",
       "      <td>0.79</td>\n",
       "      <td>490.4</td>\n",
       "      <td>1.07</td>\n",
       "      <td>7.0</td>\n",
       "      <td>20.51</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>320</td>\n",
       "      <td>44</td>\n",
       "      <td>276</td>\n",
       "      <td>2880</td>\n",
       "      <td>569</td>\n",
       "      <td>2311</td>\n",
       "      <td>0.93</td>\n",
       "      <td>0.89</td>\n",
       "      <td>551.7</td>\n",
       "      <td>1.03</td>\n",
       "      <td>4.0</td>\n",
       "      <td>19.76</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>320</td>\n",
       "      <td>44</td>\n",
       "      <td>276</td>\n",
       "      <td>3200</td>\n",
       "      <td>613</td>\n",
       "      <td>2587</td>\n",
       "      <td>1.00</td>\n",
       "      <td>1.00</td>\n",
       "      <td>613.0</td>\n",
       "      <td>1.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>19.16</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        count  actual  non_actual  cum_count  cum_actual  cum_non_actual  \\\n",
       "decile                                                                     \n",
       "0         320      80         240        320          80             240   \n",
       "1         320      89         231        640         169             471   \n",
       "2         320      86         234        960         255             705   \n",
       "3         320      75         245       1280         330             950   \n",
       "4         320      81         239       1600         411            1189   \n",
       "5         320      39         281       1920         450            1470   \n",
       "6         320      30         290       2240         480            1760   \n",
       "7         320      45         275       2560         525            2035   \n",
       "8         320      44         276       2880         569            2311   \n",
       "9         320      44         276       3200         613            2587   \n",
       "\n",
       "        percent_cum_actual  percent_cum_non_actual  if_random  lift   K_S  \\\n",
       "decile                                                                      \n",
       "0                     0.13                    0.09       61.3  1.31   4.0   \n",
       "1                     0.28                    0.18      122.6  1.38  10.0   \n",
       "2                     0.42                    0.27      183.9  1.39  15.0   \n",
       "3                     0.54                    0.37      245.2  1.35  17.0   \n",
       "4                     0.67                    0.46      306.5  1.34  21.0   \n",
       "5                     0.73                    0.57      367.8  1.22  16.0   \n",
       "6                     0.78                    0.68      429.1  1.12  10.0   \n",
       "7                     0.86                    0.79      490.4  1.07   7.0   \n",
       "8                     0.93                    0.89      551.7  1.03   4.0   \n",
       "9                     1.00                    1.00      613.0  1.00   0.0   \n",
       "\n",
       "         gain  \n",
       "decile         \n",
       "0       25.00  \n",
       "1       26.41  \n",
       "2       26.56  \n",
       "3       25.78  \n",
       "4       25.69  \n",
       "5       23.44  \n",
       "6       21.43  \n",
       "7       20.51  \n",
       "8       19.76  \n",
       "9       19.16  "
      ]
     },
     "execution_count": 519,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def createGains(model):\n",
    "    predictions = model.predict(test_hex)\n",
    "    test_scores = test_hex['loan_default'].cbind(predictions).as_data_frame()\n",
    "    #sort on prediction (descending), add id, and decile for groups containing 1/10 of datapoints\n",
    "    test_scores = test_scores.sort_values(by='predict',ascending=False)\n",
    "    test_scores['row_id'] = range(0,0+len(test_scores))\n",
    "    test_scores['decile'] = ( test_scores['row_id'] / (len(test_scores)/10) ).astype(int)\n",
    "    #see count by decile\n",
    "    test_scores.loc[test_scores['decile'] == 10]=9\n",
    "    test_scores['decile'].value_counts()\n",
    "    #create gains table\n",
    "    gains = test_scores.groupby('decile')['loan_default'].agg(['count','sum'])\n",
    "    gains.columns = ['count','actual']\n",
    "    gains\n",
    "    #add features to gains table\n",
    "    gains['non_actual'] = gains['count'] - gains['actual']\n",
    "    gains['cum_count'] = gains['count'].cumsum()\n",
    "    gains['cum_actual'] = gains['actual'].cumsum()\n",
    "    gains['cum_non_actual'] = gains['non_actual'].cumsum()\n",
    "    gains['percent_cum_actual'] = (gains['cum_actual'] / np.max(gains['cum_actual'])).round(2)\n",
    "    gains['percent_cum_non_actual'] = (gains['cum_non_actual'] / np.max(gains['cum_non_actual'])).round(2)\n",
    "    gains['if_random'] = np.max(gains['cum_actual']) /10 \n",
    "    gains['if_random'] = gains['if_random'].cumsum()\n",
    "    gains['lift'] = (gains['cum_actual'] / gains['if_random']).round(2)\n",
    "    gains['K_S'] = np.abs( gains['percent_cum_actual'] -  gains['percent_cum_non_actual'] ) * 100\n",
    "    gains['gain']=(gains['cum_actual']/gains['cum_count']*100).round(2)\n",
    "    gains = pd.DataFrame(gains)\n",
    "    return(gains)\n",
    "\n",
    "createGains(rf_tunned)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 487,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "gbm prediction progress: |████████████████████████████████████████████████| 100%\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>count</th>\n",
       "      <th>actual</th>\n",
       "      <th>non_actual</th>\n",
       "      <th>cum_count</th>\n",
       "      <th>cum_actual</th>\n",
       "      <th>cum_non_actual</th>\n",
       "      <th>percent_cum_actual</th>\n",
       "      <th>percent_cum_non_actual</th>\n",
       "      <th>if_random</th>\n",
       "      <th>lift</th>\n",
       "      <th>K_S</th>\n",
       "      <th>gain</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>decile</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>320</td>\n",
       "      <td>80</td>\n",
       "      <td>240</td>\n",
       "      <td>320</td>\n",
       "      <td>80</td>\n",
       "      <td>240</td>\n",
       "      <td>0.13</td>\n",
       "      <td>0.09</td>\n",
       "      <td>61.3</td>\n",
       "      <td>1.31</td>\n",
       "      <td>4.0</td>\n",
       "      <td>25.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>320</td>\n",
       "      <td>93</td>\n",
       "      <td>227</td>\n",
       "      <td>640</td>\n",
       "      <td>173</td>\n",
       "      <td>467</td>\n",
       "      <td>0.28</td>\n",
       "      <td>0.18</td>\n",
       "      <td>122.6</td>\n",
       "      <td>1.41</td>\n",
       "      <td>10.0</td>\n",
       "      <td>27.03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>320</td>\n",
       "      <td>102</td>\n",
       "      <td>218</td>\n",
       "      <td>960</td>\n",
       "      <td>275</td>\n",
       "      <td>685</td>\n",
       "      <td>0.45</td>\n",
       "      <td>0.26</td>\n",
       "      <td>183.9</td>\n",
       "      <td>1.50</td>\n",
       "      <td>19.0</td>\n",
       "      <td>28.65</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>320</td>\n",
       "      <td>54</td>\n",
       "      <td>266</td>\n",
       "      <td>1280</td>\n",
       "      <td>329</td>\n",
       "      <td>951</td>\n",
       "      <td>0.54</td>\n",
       "      <td>0.37</td>\n",
       "      <td>245.2</td>\n",
       "      <td>1.34</td>\n",
       "      <td>17.0</td>\n",
       "      <td>25.70</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>320</td>\n",
       "      <td>45</td>\n",
       "      <td>275</td>\n",
       "      <td>1600</td>\n",
       "      <td>374</td>\n",
       "      <td>1226</td>\n",
       "      <td>0.61</td>\n",
       "      <td>0.47</td>\n",
       "      <td>306.5</td>\n",
       "      <td>1.22</td>\n",
       "      <td>14.0</td>\n",
       "      <td>23.38</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>320</td>\n",
       "      <td>51</td>\n",
       "      <td>269</td>\n",
       "      <td>1920</td>\n",
       "      <td>425</td>\n",
       "      <td>1495</td>\n",
       "      <td>0.69</td>\n",
       "      <td>0.58</td>\n",
       "      <td>367.8</td>\n",
       "      <td>1.16</td>\n",
       "      <td>11.0</td>\n",
       "      <td>22.14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>320</td>\n",
       "      <td>51</td>\n",
       "      <td>269</td>\n",
       "      <td>2240</td>\n",
       "      <td>476</td>\n",
       "      <td>1764</td>\n",
       "      <td>0.78</td>\n",
       "      <td>0.68</td>\n",
       "      <td>429.1</td>\n",
       "      <td>1.11</td>\n",
       "      <td>10.0</td>\n",
       "      <td>21.25</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>320</td>\n",
       "      <td>44</td>\n",
       "      <td>276</td>\n",
       "      <td>2560</td>\n",
       "      <td>520</td>\n",
       "      <td>2040</td>\n",
       "      <td>0.85</td>\n",
       "      <td>0.79</td>\n",
       "      <td>490.4</td>\n",
       "      <td>1.06</td>\n",
       "      <td>6.0</td>\n",
       "      <td>20.31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>320</td>\n",
       "      <td>44</td>\n",
       "      <td>276</td>\n",
       "      <td>2880</td>\n",
       "      <td>564</td>\n",
       "      <td>2316</td>\n",
       "      <td>0.92</td>\n",
       "      <td>0.90</td>\n",
       "      <td>551.7</td>\n",
       "      <td>1.02</td>\n",
       "      <td>2.0</td>\n",
       "      <td>19.58</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>320</td>\n",
       "      <td>49</td>\n",
       "      <td>271</td>\n",
       "      <td>3200</td>\n",
       "      <td>613</td>\n",
       "      <td>2587</td>\n",
       "      <td>1.00</td>\n",
       "      <td>1.00</td>\n",
       "      <td>613.0</td>\n",
       "      <td>1.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>19.16</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        count  actual  non_actual  cum_count  cum_actual  cum_non_actual  \\\n",
       "decile                                                                     \n",
       "0         320      80         240        320          80             240   \n",
       "1         320      93         227        640         173             467   \n",
       "2         320     102         218        960         275             685   \n",
       "3         320      54         266       1280         329             951   \n",
       "4         320      45         275       1600         374            1226   \n",
       "5         320      51         269       1920         425            1495   \n",
       "6         320      51         269       2240         476            1764   \n",
       "7         320      44         276       2560         520            2040   \n",
       "8         320      44         276       2880         564            2316   \n",
       "9         320      49         271       3200         613            2587   \n",
       "\n",
       "        percent_cum_actual  percent_cum_non_actual  if_random  lift   K_S  \\\n",
       "decile                                                                      \n",
       "0                     0.13                    0.09       61.3  1.31   4.0   \n",
       "1                     0.28                    0.18      122.6  1.41  10.0   \n",
       "2                     0.45                    0.26      183.9  1.50  19.0   \n",
       "3                     0.54                    0.37      245.2  1.34  17.0   \n",
       "4                     0.61                    0.47      306.5  1.22  14.0   \n",
       "5                     0.69                    0.58      367.8  1.16  11.0   \n",
       "6                     0.78                    0.68      429.1  1.11  10.0   \n",
       "7                     0.85                    0.79      490.4  1.06   6.0   \n",
       "8                     0.92                    0.90      551.7  1.02   2.0   \n",
       "9                     1.00                    1.00      613.0  1.00   0.0   \n",
       "\n",
       "         gain  \n",
       "decile         \n",
       "0       25.00  \n",
       "1       27.03  \n",
       "2       28.65  \n",
       "3       25.70  \n",
       "4       23.38  \n",
       "5       22.14  \n",
       "6       21.25  \n",
       "7       20.31  \n",
       "8       19.58  \n",
       "9       19.16  "
      ]
     },
     "execution_count": 487,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "createGains(gbm_tunned)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 489,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stackedensemble prediction progress: |████████████████████████████████████| 100%\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>count</th>\n",
       "      <th>actual</th>\n",
       "      <th>non_actual</th>\n",
       "      <th>cum_count</th>\n",
       "      <th>cum_actual</th>\n",
       "      <th>cum_non_actual</th>\n",
       "      <th>percent_cum_actual</th>\n",
       "      <th>percent_cum_non_actual</th>\n",
       "      <th>if_random</th>\n",
       "      <th>lift</th>\n",
       "      <th>K_S</th>\n",
       "      <th>gain</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>decile</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>320</td>\n",
       "      <td>89</td>\n",
       "      <td>231</td>\n",
       "      <td>320</td>\n",
       "      <td>89</td>\n",
       "      <td>231</td>\n",
       "      <td>0.15</td>\n",
       "      <td>0.09</td>\n",
       "      <td>61.3</td>\n",
       "      <td>1.45</td>\n",
       "      <td>6.0</td>\n",
       "      <td>27.81</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>320</td>\n",
       "      <td>80</td>\n",
       "      <td>240</td>\n",
       "      <td>640</td>\n",
       "      <td>169</td>\n",
       "      <td>471</td>\n",
       "      <td>0.28</td>\n",
       "      <td>0.18</td>\n",
       "      <td>122.6</td>\n",
       "      <td>1.38</td>\n",
       "      <td>10.0</td>\n",
       "      <td>26.41</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>320</td>\n",
       "      <td>89</td>\n",
       "      <td>231</td>\n",
       "      <td>960</td>\n",
       "      <td>258</td>\n",
       "      <td>702</td>\n",
       "      <td>0.42</td>\n",
       "      <td>0.27</td>\n",
       "      <td>183.9</td>\n",
       "      <td>1.40</td>\n",
       "      <td>15.0</td>\n",
       "      <td>26.88</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>320</td>\n",
       "      <td>74</td>\n",
       "      <td>246</td>\n",
       "      <td>1280</td>\n",
       "      <td>332</td>\n",
       "      <td>948</td>\n",
       "      <td>0.54</td>\n",
       "      <td>0.37</td>\n",
       "      <td>245.2</td>\n",
       "      <td>1.35</td>\n",
       "      <td>17.0</td>\n",
       "      <td>25.94</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>320</td>\n",
       "      <td>75</td>\n",
       "      <td>245</td>\n",
       "      <td>1600</td>\n",
       "      <td>407</td>\n",
       "      <td>1193</td>\n",
       "      <td>0.66</td>\n",
       "      <td>0.46</td>\n",
       "      <td>306.5</td>\n",
       "      <td>1.33</td>\n",
       "      <td>20.0</td>\n",
       "      <td>25.44</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>320</td>\n",
       "      <td>72</td>\n",
       "      <td>248</td>\n",
       "      <td>1920</td>\n",
       "      <td>479</td>\n",
       "      <td>1441</td>\n",
       "      <td>0.78</td>\n",
       "      <td>0.56</td>\n",
       "      <td>367.8</td>\n",
       "      <td>1.30</td>\n",
       "      <td>22.0</td>\n",
       "      <td>24.95</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>320</td>\n",
       "      <td>33</td>\n",
       "      <td>287</td>\n",
       "      <td>2240</td>\n",
       "      <td>512</td>\n",
       "      <td>1728</td>\n",
       "      <td>0.84</td>\n",
       "      <td>0.67</td>\n",
       "      <td>429.1</td>\n",
       "      <td>1.19</td>\n",
       "      <td>17.0</td>\n",
       "      <td>22.86</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>320</td>\n",
       "      <td>37</td>\n",
       "      <td>283</td>\n",
       "      <td>2560</td>\n",
       "      <td>549</td>\n",
       "      <td>2011</td>\n",
       "      <td>0.90</td>\n",
       "      <td>0.78</td>\n",
       "      <td>490.4</td>\n",
       "      <td>1.12</td>\n",
       "      <td>12.0</td>\n",
       "      <td>21.45</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>320</td>\n",
       "      <td>32</td>\n",
       "      <td>288</td>\n",
       "      <td>2880</td>\n",
       "      <td>581</td>\n",
       "      <td>2299</td>\n",
       "      <td>0.95</td>\n",
       "      <td>0.89</td>\n",
       "      <td>551.7</td>\n",
       "      <td>1.05</td>\n",
       "      <td>6.0</td>\n",
       "      <td>20.17</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>320</td>\n",
       "      <td>32</td>\n",
       "      <td>288</td>\n",
       "      <td>3200</td>\n",
       "      <td>613</td>\n",
       "      <td>2587</td>\n",
       "      <td>1.00</td>\n",
       "      <td>1.00</td>\n",
       "      <td>613.0</td>\n",
       "      <td>1.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>19.16</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        count  actual  non_actual  cum_count  cum_actual  cum_non_actual  \\\n",
       "decile                                                                     \n",
       "0         320      89         231        320          89             231   \n",
       "1         320      80         240        640         169             471   \n",
       "2         320      89         231        960         258             702   \n",
       "3         320      74         246       1280         332             948   \n",
       "4         320      75         245       1600         407            1193   \n",
       "5         320      72         248       1920         479            1441   \n",
       "6         320      33         287       2240         512            1728   \n",
       "7         320      37         283       2560         549            2011   \n",
       "8         320      32         288       2880         581            2299   \n",
       "9         320      32         288       3200         613            2587   \n",
       "\n",
       "        percent_cum_actual  percent_cum_non_actual  if_random  lift   K_S  \\\n",
       "decile                                                                      \n",
       "0                     0.15                    0.09       61.3  1.45   6.0   \n",
       "1                     0.28                    0.18      122.6  1.38  10.0   \n",
       "2                     0.42                    0.27      183.9  1.40  15.0   \n",
       "3                     0.54                    0.37      245.2  1.35  17.0   \n",
       "4                     0.66                    0.46      306.5  1.33  20.0   \n",
       "5                     0.78                    0.56      367.8  1.30  22.0   \n",
       "6                     0.84                    0.67      429.1  1.19  17.0   \n",
       "7                     0.90                    0.78      490.4  1.12  12.0   \n",
       "8                     0.95                    0.89      551.7  1.05   6.0   \n",
       "9                     1.00                    1.00      613.0  1.00   0.0   \n",
       "\n",
       "         gain  \n",
       "decile         \n",
       "0       27.81  \n",
       "1       26.41  \n",
       "2       26.88  \n",
       "3       25.94  \n",
       "4       25.44  \n",
       "5       24.95  \n",
       "6       22.86  \n",
       "7       21.45  \n",
       "8       20.17  \n",
       "9       19.16  "
      ]
     },
     "execution_count": 489,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "createGains(m_SE)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As expected by averaging out some noise from different models, stacked Ensemble model performed better than base or tunned RF and GBM models. with AUC score of 0.6629, to farther improve the model I would work more on feature engineering."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Thank you!"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}