Getting Started with H2O.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**What is H2O?**<br/>\n",
    "H2O is a Java-software for data modeling anf general computing. There are many different perceptions of the H2O software, but the primary purpose of H2O is ad a distributed, parallel, in memory processing engine.\n",
    "<br/>\n",
    "\n",
    "**H2O - Key Features**<br/>\n",
    "\n",
    "Some of the key features of H2O are :\n",
    "- Access from R and Python\n",
    "- Acess from web based interface called Flow. By means of Flow, data scientist are able to import, export and modify the datasets, play with the models, verify model performance and much more.\n",
    "- AutoML : Automatic training and tuning of many models within a user specified time limit\n",
    "- Distributed, In-memory processing : In-memory processing with fast serialization between nodes and clusters to support massive datasets.\n",
    "- Simple Deployment"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import h2o\n",
    "import time\n",
    "import itertools\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from h2o.estimators.glm import H2OGeneralizedLinearEstimator\n",
    "from h2o.estimators.gbm import H2OGradientBoostingEstimator\n",
    "from h2o.estimators.random_forest import H2ORandomForestEstimator\n",
    "\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Once the module in imported, the first step is to initialize the h2o module.\n",
    "\n",
    "The h2o.init() command is pretty smart and does a lot of things. First, an attempt is made to search for an existing H2O instance being started already, before starting a new one. When none is found automatically or specified manually with argument available, a new instance of H2O is started.\n",
    "\n",
    "During startup, H2O is going to print some useful information. Version of the Python it is running on, H2O’s version, how to connect to H2O’s Flow interface or where error logs reside, just to name a few."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Checking whether there is an H2O instance running at http://localhost:54321..... not found.\n",
      "Attempting to start a local H2O server...\n",
      "  Java Version: java version \"1.8.0_181\"; Java(TM) SE Runtime Environment (build 1.8.0_181-b13); Java HotSpot(TM) 64-Bit Server VM (build 25.181-b13, mixed mode)\n",
      "  Starting server from /home/mayank/anaconda3/lib/python3.6/site-packages/h2o/backend/bin/h2o.jar\n",
      "  Ice root: /tmp/tmpwpj5u6lv\n",
      "  JVM stdout: /tmp/tmpwpj5u6lv/h2o_mayank_started_from_python.out\n",
      "  JVM stderr: /tmp/tmpwpj5u6lv/h2o_mayank_started_from_python.err\n",
      "  Server is running at http://127.0.0.1:54321\n",
      "Connecting to H2O server at http://127.0.0.1:54321... successful.\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td>H2O cluster uptime:</td>\n",
       "<td>01 secs</td></tr>\n",
       "<tr><td>H2O cluster timezone:</td>\n",
       "<td>Asia/Kolkata</td></tr>\n",
       "<tr><td>H2O data parsing timezone:</td>\n",
       "<td>UTC</td></tr>\n",
       "<tr><td>H2O cluster version:</td>\n",
       "<td>3.18.0.11</td></tr>\n",
       "<tr><td>H2O cluster version age:</td>\n",
       "<td>2 months and 20 days </td></tr>\n",
       "<tr><td>H2O cluster name:</td>\n",
       "<td>H2O_from_python_mayank_jqbcb9</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>3.455 Gb</td></tr>\n",
       "<tr><td>H2O cluster total cores:</td>\n",
       "<td>4</td></tr>\n",
       "<tr><td>H2O cluster allowed cores:</td>\n",
       "<td>4</td></tr>\n",
       "<tr><td>H2O cluster status:</td>\n",
       "<td>accepting new members, healthy</td></tr>\n",
       "<tr><td>H2O connection url:</td>\n",
       "<td>http://127.0.0.1:54321</td></tr>\n",
       "<tr><td>H2O connection proxy:</td>\n",
       "<td>None</td></tr>\n",
       "<tr><td>H2O internal security:</td>\n",
       "<td>False</td></tr>\n",
       "<tr><td>H2O API Extensions:</td>\n",
       "<td>XGBoost, Algos, AutoML, Core V3, Core V4</td></tr>\n",
       "<tr><td>Python version:</td>\n",
       "<td>3.6.4 final</td></tr></table></div>"
      ],
      "text/plain": [
       "--------------------------  ----------------------------------------\n",
       "H2O cluster uptime:         01 secs\n",
       "H2O cluster timezone:       Asia/Kolkata\n",
       "H2O data parsing timezone:  UTC\n",
       "H2O cluster version:        3.18.0.11\n",
       "H2O cluster version age:    2 months and 20 days\n",
       "H2O cluster name:           H2O_from_python_mayank_jqbcb9\n",
       "H2O cluster total nodes:    1\n",
       "H2O cluster free memory:    3.455 Gb\n",
       "H2O cluster total cores:    4\n",
       "H2O cluster allowed cores:  4\n",
       "H2O cluster status:         accepting new members, healthy\n",
       "H2O connection url:         http://127.0.0.1:54321\n",
       "H2O connection proxy:\n",
       "H2O internal security:      False\n",
       "H2O API Extensions:         XGBoost, Algos, AutoML, Core V3, Core V4\n",
       "Python version:             3.6.4 final\n",
       "--------------------------  ----------------------------------------"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "h2o.init()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Data Exploration**\n",
    "\n",
    "Now the initialization is done, let us first import the dataset. The command is very similiar to pandas.read_csv and the data is stored in memory in a data structure called H2OFrame.\n",
    "\n",
    "H2O support varies file format and data sources."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n"
     ]
    }
   ],
   "source": [
    "diabetes_df = h2o.import_file('./pima-indians-diabetes-database/diabetes.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table>\n",
       "<thead>\n",
       "<tr><th style=\"text-align: right;\">  Pregnancies</th><th style=\"text-align: right;\">  Glucose</th><th style=\"text-align: right;\">  BloodPressure</th><th style=\"text-align: right;\">  SkinThickness</th><th style=\"text-align: right;\">  Insulin</th><th style=\"text-align: right;\">  BMI</th><th style=\"text-align: right;\">  DiabetesPedigreeFunction</th><th style=\"text-align: right;\">  Age</th><th style=\"text-align: right;\">  Outcome</th></tr>\n",
       "</thead>\n",
       "<tbody>\n",
       "<tr><td style=\"text-align: right;\">            6</td><td style=\"text-align: right;\">      148</td><td style=\"text-align: right;\">             72</td><td style=\"text-align: right;\">             35</td><td style=\"text-align: right;\">        0</td><td style=\"text-align: right;\"> 33.6</td><td style=\"text-align: right;\">                     0.627</td><td style=\"text-align: right;\">   50</td><td style=\"text-align: right;\">        1</td></tr>\n",
       "<tr><td style=\"text-align: right;\">            1</td><td style=\"text-align: right;\">       85</td><td style=\"text-align: right;\">             66</td><td style=\"text-align: right;\">             29</td><td style=\"text-align: right;\">        0</td><td style=\"text-align: right;\"> 26.6</td><td style=\"text-align: right;\">                     0.351</td><td style=\"text-align: right;\">   31</td><td style=\"text-align: right;\">        0</td></tr>\n",
       "<tr><td style=\"text-align: right;\">            8</td><td style=\"text-align: right;\">      183</td><td style=\"text-align: right;\">             64</td><td style=\"text-align: right;\">              0</td><td style=\"text-align: right;\">        0</td><td style=\"text-align: right;\"> 23.3</td><td style=\"text-align: right;\">                     0.672</td><td style=\"text-align: right;\">   32</td><td style=\"text-align: right;\">        1</td></tr>\n",
       "<tr><td style=\"text-align: right;\">            1</td><td style=\"text-align: right;\">       89</td><td style=\"text-align: right;\">             66</td><td style=\"text-align: right;\">             23</td><td style=\"text-align: right;\">       94</td><td style=\"text-align: right;\"> 28.1</td><td style=\"text-align: right;\">                     0.167</td><td style=\"text-align: right;\">   21</td><td style=\"text-align: right;\">        0</td></tr>\n",
       "<tr><td style=\"text-align: right;\">            0</td><td style=\"text-align: right;\">      137</td><td style=\"text-align: right;\">             40</td><td style=\"text-align: right;\">             35</td><td style=\"text-align: right;\">      168</td><td style=\"text-align: right;\"> 43.1</td><td style=\"text-align: right;\">                     2.288</td><td style=\"text-align: right;\">   33</td><td style=\"text-align: right;\">        1</td></tr>\n",
       "<tr><td style=\"text-align: right;\">            5</td><td style=\"text-align: right;\">      116</td><td style=\"text-align: right;\">             74</td><td style=\"text-align: right;\">              0</td><td style=\"text-align: right;\">        0</td><td style=\"text-align: right;\"> 25.6</td><td style=\"text-align: right;\">                     0.201</td><td style=\"text-align: right;\">   30</td><td style=\"text-align: right;\">        0</td></tr>\n",
       "<tr><td style=\"text-align: right;\">            3</td><td style=\"text-align: right;\">       78</td><td style=\"text-align: right;\">             50</td><td style=\"text-align: right;\">             32</td><td style=\"text-align: right;\">       88</td><td style=\"text-align: right;\"> 31  </td><td style=\"text-align: right;\">                     0.248</td><td style=\"text-align: right;\">   26</td><td style=\"text-align: right;\">        1</td></tr>\n",
       "<tr><td style=\"text-align: right;\">           10</td><td style=\"text-align: right;\">      115</td><td style=\"text-align: right;\">              0</td><td style=\"text-align: right;\">              0</td><td style=\"text-align: right;\">        0</td><td style=\"text-align: right;\"> 35.3</td><td style=\"text-align: right;\">                     0.134</td><td style=\"text-align: right;\">   29</td><td style=\"text-align: right;\">        0</td></tr>\n",
       "<tr><td style=\"text-align: right;\">            2</td><td style=\"text-align: right;\">      197</td><td style=\"text-align: right;\">             70</td><td style=\"text-align: right;\">             45</td><td style=\"text-align: right;\">      543</td><td style=\"text-align: right;\"> 30.5</td><td style=\"text-align: right;\">                     0.158</td><td style=\"text-align: right;\">   53</td><td style=\"text-align: right;\">        1</td></tr>\n",
       "<tr><td style=\"text-align: right;\">            8</td><td style=\"text-align: right;\">      125</td><td style=\"text-align: right;\">             96</td><td style=\"text-align: right;\">              0</td><td style=\"text-align: right;\">        0</td><td style=\"text-align: right;\">  0  </td><td style=\"text-align: right;\">                     0.232</td><td style=\"text-align: right;\">   54</td><td style=\"text-align: right;\">        1</td></tr>\n",
       "</tbody>\n",
       "</table>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "diabetes_df.show(rows = 5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Rows:768\n",
      "Cols:9\n",
      "\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<table>\n",
       "<thead>\n",
       "<tr><th>       </th><th>Pregnancies      </th><th>Glucose           </th><th>BloodPressure     </th><th>SkinThickness     </th><th>Insulin           </th><th>BMI              </th><th>DiabetesPedigreeFunction  </th><th>Age               </th><th>Outcome            </th></tr>\n",
       "</thead>\n",
       "<tbody>\n",
       "<tr><td>type   </td><td>int              </td><td>int               </td><td>int               </td><td>int               </td><td>int               </td><td>real             </td><td>real                      </td><td>int               </td><td>int                </td></tr>\n",
       "<tr><td>mins   </td><td>0.0              </td><td>0.0               </td><td>0.0               </td><td>0.0               </td><td>0.0               </td><td>0.0              </td><td>0.078                     </td><td>21.0              </td><td>0.0                </td></tr>\n",
       "<tr><td>mean   </td><td>3.845052083333336</td><td>120.89453125      </td><td>69.10546874999994 </td><td>20.536458333333357</td><td>79.79947916666666 </td><td>31.99257812500003</td><td>0.4718763020833334        </td><td>33.240885416666615</td><td>0.3489583333333333 </td></tr>\n",
       "<tr><td>maxs   </td><td>17.0             </td><td>199.0             </td><td>122.0             </td><td>99.0              </td><td>846.0             </td><td>67.1             </td><td>2.42                      </td><td>81.0              </td><td>1.0                </td></tr>\n",
       "<tr><td>sigma  </td><td>3.36957806269887 </td><td>31.972618195136224</td><td>19.355807170644777</td><td>15.952217567727642</td><td>115.24400235133803</td><td>7.884160320375441</td><td>0.331328595012775         </td><td>11.760231540678689</td><td>0.47695137724279896</td></tr>\n",
       "<tr><td>zeros  </td><td>111              </td><td>5                 </td><td>35                </td><td>227               </td><td>374               </td><td>11               </td><td>0                         </td><td>0                 </td><td>500                </td></tr>\n",
       "<tr><td>missing</td><td>0                </td><td>0                 </td><td>0                 </td><td>0                 </td><td>0                 </td><td>0                </td><td>0                         </td><td>0                 </td><td>0                  </td></tr>\n",
       "<tr><td>0      </td><td>6.0              </td><td>148.0             </td><td>72.0              </td><td>35.0              </td><td>0.0               </td><td>33.6             </td><td>0.627                     </td><td>50.0              </td><td>1.0                </td></tr>\n",
       "<tr><td>1      </td><td>1.0              </td><td>85.0              </td><td>66.0              </td><td>29.0              </td><td>0.0               </td><td>26.6             </td><td>0.351                     </td><td>31.0              </td><td>0.0                </td></tr>\n",
       "<tr><td>2      </td><td>8.0              </td><td>183.0             </td><td>64.0              </td><td>0.0               </td><td>0.0               </td><td>23.3             </td><td>0.672                     </td><td>32.0              </td><td>1.0                </td></tr>\n",
       "<tr><td>3      </td><td>1.0              </td><td>89.0              </td><td>66.0              </td><td>23.0              </td><td>94.0              </td><td>28.1             </td><td>0.167                     </td><td>21.0              </td><td>0.0                </td></tr>\n",
       "<tr><td>4      </td><td>0.0              </td><td>137.0             </td><td>40.0              </td><td>35.0              </td><td>168.0             </td><td>43.1             </td><td>2.288                     </td><td>33.0              </td><td>1.0                </td></tr>\n",
       "<tr><td>5      </td><td>5.0              </td><td>116.0             </td><td>74.0              </td><td>0.0               </td><td>0.0               </td><td>25.6             </td><td>0.201                     </td><td>30.0              </td><td>0.0                </td></tr>\n",
       "<tr><td>6      </td><td>3.0              </td><td>78.0              </td><td>50.0              </td><td>32.0              </td><td>88.0              </td><td>31.0             </td><td>0.248                     </td><td>26.0              </td><td>1.0                </td></tr>\n",
       "<tr><td>7      </td><td>10.0             </td><td>115.0             </td><td>0.0               </td><td>0.0               </td><td>0.0               </td><td>35.3             </td><td>0.134                     </td><td>29.0              </td><td>0.0                </td></tr>\n",
       "<tr><td>8      </td><td>2.0              </td><td>197.0             </td><td>70.0              </td><td>45.0              </td><td>543.0             </td><td>30.5             </td><td>0.158                     </td><td>53.0              </td><td>1.0                </td></tr>\n",
       "<tr><td>9      </td><td>8.0              </td><td>125.0             </td><td>96.0              </td><td>0.0               </td><td>0.0               </td><td>0.0              </td><td>0.232                     </td><td>54.0              </td><td>1.0                </td></tr>\n",
       "</tbody>\n",
       "</table>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "diabetes_df.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "describe() gives out a lot of information.\n",
    "\n",
    "- Number of rows and columns in the dataset\n",
    "- A number of summary statistics about the dataset such as\n",
    "- Data type of the column such as integer, categorical etc\n",
    "- Minimum value\n",
    "- Mean value\n",
    "- Maximum value\n",
    "- Standard deviation value\n",
    "- Number of zeros in the column\n",
    "- Number of missing values in the column\n",
    "- A look at the top few rows\n",
    "\n",
    "\n",
    "Now let us look at the distribution of the individual features using hist() command"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mayank/anaconda3/lib/python3.6/site-packages/matplotlib/__init__.py:1717: MatplotlibDeprecationWarning: The *left* kwarg to `bar` is deprecated use *x* instead. Support for *left* will be removed in Matplotlib 3.0\n",
      "  return func(ax, *args, **kwargs)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fd07cbec4a8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fd07c9494a8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fd07c8ca828>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fd07a891b00>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fd07a7840f0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fd07c966908>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fd07a757cf8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fd07a7dbef0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEWCAYAAAB8LwAVAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvNQv5yAAAGPpJREFUeJzt3Xu0XGWd5vHvAwERuYmJiCEYlTgKjlw6zeDY4wVsF2ILaIsNg3KZKLME7VZ7bNFxWvpiD8yoNCx7lNg4AqJysQW8tSKKtDaoQe5eGkSECANRLgEBEfjNH/s9UiQ7OXVC6tQh+X7WqnX2fuutXb/aK6mn9rur3p2qQpKkFW0w7gIkSTOTASFJ6mVASJJ6GRCSpF4GhCSplwEhSeplQGhaJbkmyUvHXcc4JXlNkpuS3JNk13HXI62KAaG1JskNSV6+QtthSb49sV5VO1XVhZNsZ36SSjJrRKWO2weBt1bVZlV12Yp3pvOuJNcmuS/JjUmOTfKEYZ+g7b8d1mrVWu8YEFrvzIDgeQZwzWruPxE4AjgE2Bx4JbAncOboS5MeYUBoWg0eZSTZPcmSJMuT3Jrkw63bRe3vnW0Y5oVJNkjyviQ/T3JbklOTbDmw3UPafb9K8j9WeJ5jkpyd5FNJlgOHtee+OMmdSW5J8pEkGw9sr5Ic2T7F353kb5I8uz1meZIzB/uv8Bp7a03yhCT3ABsCVyT5ac9jFwBHAgdX1cVV9WBVXQP8MbB3kj1bvwuTvGngcb87Uksysf+uaPvvT1r7fkkub/X/NMnerf3pSc5LcnuS65K8eWC7xyQ5q+27u5NcleQ5Sd7TXttNSV4x0H/LJCe3ffqLJH+bZMPJ/2VoJjIgNE4nACdU1RbAs3nkE/KL29+t2jDMxcBh7fYy4FnAZsBHAJLsCPwf4GBgW2BLYO4Kz7UfcDawFXA68BDwDmA28EJgL7o35kF7A78H7AH8BbC4Pcc84PnAQat4Xb21VtVvqmqz1mfnqnp2z2P3ApZW1fcGG6vqJuAS4A9X8ZyDfSf2385t/52RZHfgVOBddPvgxcANrd9ngKXA04HXAX+XZK+BTb4aOA14MnAZ8FW69465wF8DJw30PQV4ENgB2BV4BfAm9LhkQGhtO6d9Kr8zyZ10b9yr8ltghySzq+qeqrpkNX0PBj5cVddX1T3Ae4AD23DR64AvVNW3q+oB4C+BFScZu7iqzqmqh6vqvqq6tKouaZ/Qb6B7k3vJCo85rqqWt0/wVwNfa89/F/AVujfAqdY6mdnALau475Z2/5pYBHyiqs5v++AXVfXjJPOAPwDeXVX3V9XlwD8Cbxx47L9U1Ver6kHgLGAOcGxV/Rb4LDA/yVZJtqEbDnt7Vf26qm4DjgcOXMOaNWYGhNa2/atqq4kbK38qH7QIeA7w4yTfT/JHq+n7dODnA+s/B2YB27T7bpq4o6ruBX61wuNvGlxpwyRfTPL/2rDT37Hym++tA8v39axvRr/V1TqZX9IdBfXZtt2/JuYBKw1p0dV6e1XdPdD2cx59BLbi6/5lVT00sA7dvngGsBFwy8AHhJOAp65hzRozA0JjU1XXVtVBdG8gxwFnJ3kSK3/6B7iZ7g1owvZ0Qxm30n2y3m7ijiRPBJ6y4tOtsP5R4MfAgjbE9V4ga/5qhq51Mt8A5rUhod9pn/T3AC5oTb8GNh3o8rRJtnsT3TBeX61bJ9l8hXp/MUStfc/xG2D2wIeELapqpzXYlmYAA0Jjk+QNSeZU1cPAna35IWAZ8DDd+P2EzwDvSPLMJJvRfeI/ow17nA28Osl/bCeO/4rJ3+w3B5YD9yR5LvCWtfbCVl/ralXVvwEfA05PskeSDZPsBHwO+HpVfb11vRx4bZJN032dddEKm7qVR++/k4HDk+zVTqLPTfLcdm7jX4H/mWSTJC9o2zp9qi+6qm4BvgZ8KMkW7XmenWTFoTs9ThgQGqe9gWvaN3tOAA5s4+D3Ah8AvtOGKvYAPkF3ovQi4GfA/cDbANo5grfRjYffAtwN3Eb3aXZV/hvwn1vfjwNnrMXXtcpah/RWuvMAnwLuAf4ZuJDum0wTjgceoAuCU1j5Df0Y4JS2/17fTnof3h53F/AtHjnKOQiYT3c08Xng/VV1/hTqHXQIsDHwQ+AOuvBe1ZCZZrh4wSCta9qn9jvpho9+Nu56pMcrjyC0Tkjy6jbc8iS6XypfxSNf45S0BgwIrSv2oxsiuRlYQDdc5eGx9Bg4xCRJ6uURhCSp17gnLXtMZs+eXfPnzx93GZL0uHLppZf+sqrmTNbvcR0Q8+fPZ8mSJeMuQ5IeV5L8fPJeDjFJklbBgJAk9TIgJEm9DAhJUi8DQpLUy4CQJPUaaUCkuy7wVe06uEta29ZJzm/X+j0/yZNbe5Kc2K6Je2WS3UZZmyRp9abjCOJlVbVLVS1s60cDF1TVArqLnxzd2l9JN4fOAuAIugu6SJLGZBxDTPvRzV9P+7v/QPup1bkE2CqJ88hL0piM+pfUBXwtSQEnVdViYJt25Smq6pYkE9erncujrxu8tLU96gLuSY6gO8Jg++23X+PC5h/9pTV+7GN1w7GvGttzS9KwRh0QL6qqm1sInJ/kx6vp23eJyJWmmm0hsxhg4cKFTkUrSSMy0iGmqrq5/b2N7lKGuwO3Tgwdtb+3te5LgXkDD9+Obm5/SdIYjCwgkjwpyeYTy8ArgKuB84BDW7dDgXPb8nnAIe3bTHsAd00MRUmSpt8oh5i2AT6fZOJ5Pl1V/5zk+8CZSRYBNwIHtP5fBvYBrgPupbvAuiRpTEYWEFV1PbBzT/uvgL162gs4alT1SJKmxl9SS5J6GRCSpF4GhCSplwEhSeplQEiSehkQkqReBoQkqZcBIUnqZUBIknoZEJKkXgaEJKmXASFJ6mVASJJ6GRCSpF4GhCSplwEhSeplQEiSehkQkqReBoQkqZcBIUnqZUBIknoZEJKkXgaEJKmXASFJ6mVASJJ6GRCSpF4GhCSplwEhSeplQEiSehkQkqReBoQkqZcBIUnqNfKASLJhksuSfLGtPzPJd5Ncm+SMJBu39ie09eva/fNHXZskadWm4wjiz4AfDawfBxxfVQuAO4BFrX0RcEdV7QAc3/pJksZkpAGRZDvgVcA/tvUAewJnty6nAPu35f3aOu3+vVp/SdIYjPoI4u+BvwAebutPAe6sqgfb+lJgblueC9wE0O6/q/V/lCRHJFmSZMmyZctGWbskrddGFhBJ/gi4raouHWzu6VpD3PdIQ9XiqlpYVQvnzJmzFiqVJPWZNcJtvwjYN8k+wCbAFnRHFFslmdWOErYDbm79lwLzgKVJZgFbArePsD5J0mqM7Aiiqt5TVdtV1XzgQOAbVXUw8E3gda3bocC5bfm8tk67/xtVtdIRhCRpeozjdxDvBt6Z5Dq6cwwnt/aTgae09ncCR4+hNklSM8ohpt+pqguBC9vy9cDuPX3uBw6YjnokSZPzl9SSpF4GhCSplwEhSeplQEiSehkQkqReBoQkqZcBIUnqZUBIknoZEJKkXgaEJKmXASFJ6mVASJJ6GRCSpF4GhCSplwEhSeplQEiSehkQkqReBoQkqZcBIUnqZUBIknoZEJKkXgaEJKmXASFJ6mVASJJ6GRCSpF4GhCSplwEhSeo1VEAkef6oC5EkzSzDHkF8LMn3khyZZKuRViRJmhGGCoiq+gPgYGAesCTJp5P84UgrkySN1dDnIKrqWuB9wLuBlwAnJvlxkteOqjhJ0vgMew7iBUmOB34E7Am8uqqe15aPH2F9kqQxGfYI4iPAD4Cdq+qoqvoBQFXdTHdUsZIkm7TzFlckuSbJX7X2Zyb5bpJrk5yRZOPW/oS2fl27f/5jfXGSpDU3bEDsA3y6qu4DSLJBkk0Bquq0VTzmN8CeVbUzsAuwd5I9gOOA46tqAXAHsKj1XwTcUVU70B2VHLcmL0iStHYMGxBfB544sL5pa1ul6tzTVjdqt6Ibljq7tZ8C7N+W92vrtPv3SpIh65MkrWXDBsQmA2/2tOVNJ3tQkg2TXA7cBpwP/BS4s6oebF2WAnPb8lzgprb9B4G7gKcMWZ8kaS0bNiB+nWS3iZUkvwfcN9mDquqhqtoF2A7YHXheX7eJza7mvt9JckSSJUmWLFu2bKjiJUlTN2vIfm8Hzkpyc1vfFviTYZ+kqu5MciGwB7BVklntKGE7YGKbS+l+Z7E0ySxgS+D2nm0tBhYDLFy4cKUAkSStHcP+UO77wHOBtwBHAs+rqktX95gkcyZ+dZ3kicDL6b4m+03gda3bocC5bfm8tk67/xtVZQBI0pgMewQB8PvA/PaYXZNQVaeupv+2wClJNqQLojOr6otJfgh8NsnfApcBJ7f+JwOnJbmO7sjhwKm9FEnS2jRUQCQ5DXg2cDnwUGsuYJUBUVVXArv2tF9Pdz5ixfb7gQOGqUeSNHrDHkEsBHZ0yEeS1h/DfovpauBpoyxEkjSzDHsEMRv4YZLv0f1CGoCq2nckVUmSxm7YgDhmlEVIkmaeoQKiqr6V5BnAgqr6epuHacPRliZJGqdhp/t+M938SCe1prnAOaMqSpI0fsOepD4KeBGwHH538aCnjqooSdL4DRsQv6mqByZW2lQYfuVVktZhw56k/laS9wJPbNeiPhL4wujKkqSZb/7RXxrbc99w7KtG/hzDHkEcDSwDrgL+K/BlVnElOUnSumHYbzE9DHy83SRJ64Fh52L6GT3nHKrqWWu9IknSjDCVuZgmbEI3qd7Wa78cSdJMMez1IH41cPtFVf093bWlJUnrqGGHmHYbWN2A7ohi85FUJEmaEYYdYvrQwPKDwA3A69d6NZKkGWPYbzG9bNSFSJJmlmGHmN65uvur6sNrpxxJ0kwxlW8x/T5wXlt/NXARcNMoipIkjd9ULhi0W1XdDZDkGOCsqnrTqAqTJI3XsFNtbA88MLD+ADB/rVcjSZoxhj2COA34XpLP0/2i+jXAqSOrSpI0dsN+i+kDSb4C/KfWdHhVXTa6siRJ4zbsEBPApsDyqjoBWJrkmSOqSZI0Awx7ydH3A+8G3tOaNgI+NaqiJEnjN+wRxGuAfYFfA1TVzTjVhiSt04YNiAeqqmhTfid50uhKkiTNBMMGxJlJTgK2SvJm4Ot48SBJWqcN+y2mD7ZrUS8H/h3wl1V1/kgrkySN1aQBkWRD4KtV9XLAUJCk9cSkQ0xV9RBwb5Itp6EeSdIMMewvqe8HrkpyPu2bTABV9acjqUqSNHbDBsSX2k2StJ5YbUAk2b6qbqyqU6a64STz6OZrehrwMLC4qk5IsjVwBt1kfzcAr6+qO5IEOAHYB7gXOKyqfjDV55UkrR2TnYM4Z2IhyeemuO0HgT+vqucBewBHJdkROBq4oKoWABe0dYBXAgva7Qjgo1N8PknSWjRZQGRg+VlT2XBV3TJxBNCuI/EjYC6wHzBxRHIKsH9b3g84tTqX0P3mYtupPKckae2ZLCBqFctTkmQ+sCvwXWCbqroFuhABntq6zeXRV6hb2tpW3NYRSZYkWbJs2bI1LUmSNInJAmLnJMuT3A28oC0vT3J3kuXDPEGSzYDPAW+vqtU9Jj1tK4VSVS2uqoVVtXDOnDnDlCBJWgOrPUldVRs+lo0n2YguHE6vqn9qzbcm2baqbmlDSLe19qXAvIGHbwfc/FieX5K05qZyPYgpad9KOhn4UVV9eOCu84BD2/KhwLkD7Yekswdw18RQlCRp+g37O4g18SLgjXQ/sLu8tb0XOJZu8r9FwI3AAe2+L9N9xfU6uq+5Hj7C2iRJkxhZQFTVt+k/rwCwV0//Ao4aVT2SpKkZ2RCTJOnxzYCQJPUyICRJvQwISVIvA0KS1MuAkCT1MiAkSb0MCElSLwNCktTLgJAk9TIgJEm9DAhJUi8DQpLUy4CQJPUyICRJvQwISVIvA0KS1MuAkCT1MiAkSb0MCElSLwNCktTLgJAk9TIgJEm9DAhJUi8DQpLUy4CQJPUyICRJvQwISVIvA0KS1MuAkCT1MiAkSb0MCElSr5EFRJJPJLktydUDbVsnOT/Jte3vk1t7kpyY5LokVybZbVR1SZKGM8ojiE8Ce6/QdjRwQVUtAC5o6wCvBBa02xHAR0dYlyRpCCMLiKq6CLh9heb9gFPa8inA/gPtp1bnEmCrJNuOqjZJ0uSm+xzENlV1C0D7+9TWPhe4aaDf0ta2kiRHJFmSZMmyZctGWqwkrc9myknq9LRVX8eqWlxVC6tq4Zw5c0ZcliStv6Y7IG6dGDpqf29r7UuBeQP9tgNunubaJEkDpjsgzgMObcuHAucOtB/Svs20B3DXxFCUJGk8Zo1qw0k+A7wUmJ1kKfB+4FjgzCSLgBuBA1r3LwP7ANcB9wKHj6ouSdJwRhYQVXXQKu7aq6dvAUeNqhZJ0tTNlJPUkqQZxoCQJPUyICRJvQwISVIvA0KS1MuAkCT1MiAkSb0MCElSLwNCktTLgJAk9TIgJEm9DAhJUi8DQpLUy4CQJPUyICRJvQwISVIvA0KS1MuAkCT1MiAkSb0MCElSLwNCktTLgJAk9TIgJEm9DAhJUi8DQpLUy4CQJPUyICRJvQwISVIvA0KS1MuAkCT1MiAkSb0MCElSLwNCktRrRgVEkr2T/CTJdUmOHnc9krQ+mzEBkWRD4B+AVwI7Agcl2XG8VUnS+mvGBASwO3BdVV1fVQ8AnwX2G3NNkrTemjXuAgbMBW4aWF8K/IcVOyU5Ajiird6T5CfTUNtUzAZ+uboOOW6aKhm/SffFesR90XE/POIx7YvH+D7yjGE6zaSASE9brdRQtRhYPPpy1kySJVW1cNx1zATui0e4Lzruh0c8HvbFTBpiWgrMG1jfDrh5TLVI0npvJgXE94EFSZ6ZZGPgQOC8MdckSeutGTPEVFUPJnkr8FVgQ+ATVXXNmMtaEzN2+GsM3BePcF903A+PmPH7IlUrDfNLkjSjhpgkSTOIASFJ6mVArIHJpgRJ8s4kP0xyZZILkgz1nePHo2GnR0nyuiSVZEZ/re+xGGZfJHl9+7dxTZJPT3eN02WI/yPbJ/lmksva/5N9xlHndEjyiSS3Jbl6FfcnyYltX12ZZLfprnGVqsrbFG50J9B/CjwL2Bi4AthxhT4vAzZty28Bzhh33ePaF63f5sBFwCXAwnHXPcZ/FwuAy4Ant/WnjrvuMe6LxcBb2vKOwA3jrnuE++PFwG7A1au4fx/gK3S/BdsD+O64a564eQQxdZNOCVJV36yqe9vqJXS/6VgXDTs9yt8A/wu4fzqLm2bD7Is3A/9QVXcAVNVt01zjdBlmXxSwRVveknX4N09VdRFw+2q67AecWp1LgK2SbDs91a2eATF1fVOCzF1N/0V0nw7WRZPuiyS7AvOq6ovTWdgYDPPv4jnAc5J8J8klSfaetuqm1zD74hjgDUmWAl8G3jY9pc1IU31PmTYz5ncQjyNDTQkCkOQNwELgJSOtaHxWuy+SbAAcDxw2XQWN0TD/LmbRDTO9lO6o8l+SPL+q7hxxbdNtmH1xEPDJqvpQkhcCp7V98fDoy5txhn5PmW4eQUzdUFOCJHk58N+BfavqN9NU23SbbF9sDjwfuDDJDXTjq+etoyeqh/l3sRQ4t6p+W1U/A35CFxjrmmH2xSLgTICquhjYhG7yuvXRjJ1myICYukmnBGnDKifRhcO6Os4Mk+yLqrqrqmZX1fyqmk93PmbfqloynnJHapipYs6h+wIDSWbTDTldP61VTo9h9sWNwF4ASZ5HFxDLprXKmeM84JD2baY9gLuq6pZxFwUOMU1ZrWJKkCR/DSypqvOA/w1sBpyVBODGqtp3bEWPyJD7Yr0w5L74KvCKJD8EHgLeVVW/Gl/VozHkvvhz4ONJ3kE3nHJYta/0rGuSfIZuWHF2O+fyfmAjgKr6GN05mH2A64B7gcPHU+nKnGpDktTLISZJUi8DQpLUy4CQJPUyICRJvQwISVIvA0LrvSTbJTk3ybVJfprkhPb9/dU95r3TVZ80LgaE1mvpfqjyT8A5VbWA7sdrmwEfmOShBoTWeQaE1nd7AvdX1f8FqKqHgHcA/yXJkUk+MtExyReTvDTJscATk1ye5PR23yFtLv8rkpzW2p7RrgcycV2Q7Vv7J5N8tF0P4fokL2nXDPhRkk8OPN8rklyc5AdJzkqy2bTtFQkDQtoJuHSwoaqW000F0TvTQFUdDdxXVbtU1cFJdqKbd2vPqtoZ+LPW9SN00zi/ADgdOHFgM0+mC6d3AF+gm9RwJ+DfJ9mlTcXxPuDlVbUbsAR459p4wdKwnGpD67vQP3Pmqtr77AmcXVW/BKiqibn/Xwi8ti2fRndNjAlfqKpKchVwa1VdBZDkGmA+3YRtOwLfadO1bAxcPGQ90lphQGh9dw3wx4MNSbagm13zLh59lL3JKrYxbJgM9pmY4ffhgeWJ9Vl0czWdX1UHDbFdaSQcYtL67gJg0ySHACTZEPgQ8Em6mVZ3SbJBknl0V0qb8NskGw1s4/VJntK2sXVr/1e6mUwBDga+PYW6LgFelGSHts1Nkzxnqi9OeiwMCK3X2gyirwEOSHIt8G90l0Z9L/Ad4GfAVcAHgR8MPHQxcGWS06vqGrpvPX0ryRXAh1ufPwUOT3Il8EYeOTcxTF3L6C609Jn2+EuA567p65TWhLO5SpJ6eQQhSeplQEiSehkQkqReBoQkqZcBIUnqZUBIknoZEJKkXv8fFdt2aJ65u84AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fd07a757a20>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for col in diabetes_df.columns:\n",
    "    diabetes_df[col].hist()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now let us also look at the correlation of the individual features. We can use the cor() function in H2OFrame for the same."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fd07a7ac978>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize = (10, 10))\n",
    "corr = diabetes_df.cor().as_data_frame()\n",
    "corr.index = diabetes_df.columns\n",
    "\n",
    "sns.heatmap(data = corr, annot = True, cmap='RdBu_r', vmax = 1, vmin = -1)\n",
    "\n",
    "plt.title('Correlation Heatmap', fontsize = 16)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let us now split the data into three parts - train, valid and test datasets - at a ratio of 60%, 20% and 20% respectively. We could use `split_frame()` function for the same."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "train, val, test = diabetes_df.split_frame(ratios = [0.6, 0.2], seed = 100)\n",
    "\n",
    "response = \"Outcome\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "----------------------------------------------------------------------------------------------------\n",
      "Number of Rows in Train : 459, Test : 166 and Valid : 143\n"
     ]
    }
   ],
   "source": [
    "train[response] = train[response].asfactor()\n",
    "test[response] = test[response].asfactor()\n",
    "val[response] = val[response].asfactor()\n",
    "\n",
    "print('-'*100)\n",
    "print('Number of Rows in Train : {}, Test : {} and Valid : {}'.format(train.shape[0], test.shape[0], val.shape[0]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Modeling and Model Training**\n",
    "\n",
    "Now let us built a basline model using these three splits."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "gbm Model Build progress: |███████████████████████████████████████████████| 100%\n"
     ]
    }
   ],
   "source": [
    "feature_variables = diabetes_df.columns[:-1]\n",
    "gbm = H2OGradientBoostingEstimator()\n",
    "gbm.train(x = feature_variables, y = response, training_frame = train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model Details\n",
      "=============\n",
      "H2OGradientBoostingEstimator :  Gradient Boosting Machine\n",
      "Model Key:  GBM_model_python_1534163112715_1\n",
      "\n",
      "\n",
      "ModelMetricsBinomial: gbm\n",
      "** Reported on train data. **\n",
      "\n",
      "MSE: 0.045362673251920275\n",
      "RMSE: 0.21298514796088547\n",
      "LogLoss: 0.19003856764530652\n",
      "Mean Per-Class Error: 0.029322162985529365\n",
      "AUC: 0.9959274773631209\n",
      "Gini: 0.9918549547262419\n",
      "Confusion Matrix (Act/Pred) for max f1 @ threshold = 0.39428446288510066: \n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b></b></td>\n",
       "<td><b>0</b></td>\n",
       "<td><b>1</b></td>\n",
       "<td><b>Error</b></td>\n",
       "<td><b>Rate</b></td></tr>\n",
       "<tr><td>0</td>\n",
       "<td>293.0</td>\n",
       "<td>10.0</td>\n",
       "<td>0.033</td>\n",
       "<td> (10.0/303.0)</td></tr>\n",
       "<tr><td>1</td>\n",
       "<td>4.0</td>\n",
       "<td>152.0</td>\n",
       "<td>0.0256</td>\n",
       "<td> (4.0/156.0)</td></tr>\n",
       "<tr><td>Total</td>\n",
       "<td>297.0</td>\n",
       "<td>162.0</td>\n",
       "<td>0.0305</td>\n",
       "<td> (14.0/459.0)</td></tr></table></div>"
      ],
      "text/plain": [
       "       0    1    Error    Rate\n",
       "-----  ---  ---  -------  ------------\n",
       "0      293  10   0.033    (10.0/303.0)\n",
       "1      4    152  0.0256   (4.0/156.0)\n",
       "Total  297  162  0.0305   (14.0/459.0)"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Maximum Metrics: Maximum metrics at their respective thresholds\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b>metric</b></td>\n",
       "<td><b>threshold</b></td>\n",
       "<td><b>value</b></td>\n",
       "<td><b>idx</b></td></tr>\n",
       "<tr><td>max f1</td>\n",
       "<td>0.3942845</td>\n",
       "<td>0.9559748</td>\n",
       "<td>153.0</td></tr>\n",
       "<tr><td>max f2</td>\n",
       "<td>0.3362619</td>\n",
       "<td>0.9675406</td>\n",
       "<td>167.0</td></tr>\n",
       "<tr><td>max f0point5</td>\n",
       "<td>0.5265908</td>\n",
       "<td>0.9737569</td>\n",
       "<td>134.0</td></tr>\n",
       "<tr><td>max accuracy</td>\n",
       "<td>0.3942845</td>\n",
       "<td>0.9694989</td>\n",
       "<td>153.0</td></tr>\n",
       "<tr><td>max precision</td>\n",
       "<td>0.9704768</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td></tr>\n",
       "<tr><td>max recall</td>\n",
       "<td>0.2535496</td>\n",
       "<td>1.0</td>\n",
       "<td>186.0</td></tr>\n",
       "<tr><td>max specificity</td>\n",
       "<td>0.9704768</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td></tr>\n",
       "<tr><td>max absolute_mcc</td>\n",
       "<td>0.3942845</td>\n",
       "<td>0.9330429</td>\n",
       "<td>153.0</td></tr>\n",
       "<tr><td>max min_per_class_accuracy</td>\n",
       "<td>0.4253254</td>\n",
       "<td>0.9669967</td>\n",
       "<td>152.0</td></tr>\n",
       "<tr><td>max mean_per_class_accuracy</td>\n",
       "<td>0.3942845</td>\n",
       "<td>0.9706778</td>\n",
       "<td>153.0</td></tr></table></div>"
      ],
      "text/plain": [
       "metric                       threshold    value     idx\n",
       "---------------------------  -----------  --------  -----\n",
       "max f1                       0.394284     0.955975  153\n",
       "max f2                       0.336262     0.967541  167\n",
       "max f0point5                 0.526591     0.973757  134\n",
       "max accuracy                 0.394284     0.969499  153\n",
       "max precision                0.970477     1         0\n",
       "max recall                   0.25355      1         186\n",
       "max specificity              0.970477     1         0\n",
       "max absolute_mcc             0.394284     0.933043  153\n",
       "max min_per_class_accuracy   0.425325     0.966997  152\n",
       "max mean_per_class_accuracy  0.394284     0.970678  153"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Gains/Lift Table: Avg response rate: 33.99 %\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b></b></td>\n",
       "<td><b>group</b></td>\n",
       "<td><b>cumulative_data_fraction</b></td>\n",
       "<td><b>lower_threshold</b></td>\n",
       "<td><b>lift</b></td>\n",
       "<td><b>cumulative_lift</b></td>\n",
       "<td><b>response_rate</b></td>\n",
       "<td><b>cumulative_response_rate</b></td>\n",
       "<td><b>capture_rate</b></td>\n",
       "<td><b>cumulative_capture_rate</b></td>\n",
       "<td><b>gain</b></td>\n",
       "<td><b>cumulative_gain</b></td></tr>\n",
       "<tr><td></td>\n",
       "<td>1</td>\n",
       "<td>0.0108932</td>\n",
       "<td>0.9567279</td>\n",
       "<td>2.9423077</td>\n",
       "<td>2.9423077</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0320513</td>\n",
       "<td>0.0320513</td>\n",
       "<td>194.2307692</td>\n",
       "<td>194.2307692</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2</td>\n",
       "<td>0.0217865</td>\n",
       "<td>0.9504769</td>\n",
       "<td>2.9423077</td>\n",
       "<td>2.9423077</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0320513</td>\n",
       "<td>0.0641026</td>\n",
       "<td>194.2307692</td>\n",
       "<td>194.2307692</td></tr>\n",
       "<tr><td></td>\n",
       "<td>3</td>\n",
       "<td>0.0305011</td>\n",
       "<td>0.9392770</td>\n",
       "<td>2.9423077</td>\n",
       "<td>2.9423077</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0256410</td>\n",
       "<td>0.0897436</td>\n",
       "<td>194.2307692</td>\n",
       "<td>194.2307692</td></tr>\n",
       "<tr><td></td>\n",
       "<td>4</td>\n",
       "<td>0.0413943</td>\n",
       "<td>0.9289064</td>\n",
       "<td>2.9423077</td>\n",
       "<td>2.9423077</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0320513</td>\n",
       "<td>0.1217949</td>\n",
       "<td>194.2307692</td>\n",
       "<td>194.2307692</td></tr>\n",
       "<tr><td></td>\n",
       "<td>5</td>\n",
       "<td>0.0501089</td>\n",
       "<td>0.9199767</td>\n",
       "<td>2.9423077</td>\n",
       "<td>2.9423077</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0256410</td>\n",
       "<td>0.1474359</td>\n",
       "<td>194.2307692</td>\n",
       "<td>194.2307692</td></tr>\n",
       "<tr><td></td>\n",
       "<td>6</td>\n",
       "<td>0.1002179</td>\n",
       "<td>0.8906729</td>\n",
       "<td>2.9423077</td>\n",
       "<td>2.9423077</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.1474359</td>\n",
       "<td>0.2948718</td>\n",
       "<td>194.2307692</td>\n",
       "<td>194.2307692</td></tr>\n",
       "<tr><td></td>\n",
       "<td>7</td>\n",
       "<td>0.1503268</td>\n",
       "<td>0.8593380</td>\n",
       "<td>2.9423077</td>\n",
       "<td>2.9423077</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.1474359</td>\n",
       "<td>0.4423077</td>\n",
       "<td>194.2307692</td>\n",
       "<td>194.2307692</td></tr>\n",
       "<tr><td></td>\n",
       "<td>8</td>\n",
       "<td>0.2004357</td>\n",
       "<td>0.7616949</td>\n",
       "<td>2.9423077</td>\n",
       "<td>2.9423077</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.1474359</td>\n",
       "<td>0.5897436</td>\n",
       "<td>194.2307692</td>\n",
       "<td>194.2307692</td></tr>\n",
       "<tr><td></td>\n",
       "<td>9</td>\n",
       "<td>0.3006536</td>\n",
       "<td>0.5366458</td>\n",
       "<td>2.8783445</td>\n",
       "<td>2.9209866</td>\n",
       "<td>0.9782609</td>\n",
       "<td>0.9927536</td>\n",
       "<td>0.2884615</td>\n",
       "<td>0.8782051</td>\n",
       "<td>187.8344482</td>\n",
       "<td>192.0986622</td></tr>\n",
       "<tr><td></td>\n",
       "<td>10</td>\n",
       "<td>0.4008715</td>\n",
       "<td>0.2947199</td>\n",
       "<td>1.1513378</td>\n",
       "<td>2.4785744</td>\n",
       "<td>0.3913043</td>\n",
       "<td>0.8423913</td>\n",
       "<td>0.1153846</td>\n",
       "<td>0.9935897</td>\n",
       "<td>15.1337793</td>\n",
       "<td>147.8574415</td></tr>\n",
       "<tr><td></td>\n",
       "<td>11</td>\n",
       "<td>0.5010893</td>\n",
       "<td>0.1778608</td>\n",
       "<td>0.0639632</td>\n",
       "<td>1.9956522</td>\n",
       "<td>0.0217391</td>\n",
       "<td>0.6782609</td>\n",
       "<td>0.0064103</td>\n",
       "<td>1.0</td>\n",
       "<td>-93.6036789</td>\n",
       "<td>99.5652174</td></tr>\n",
       "<tr><td></td>\n",
       "<td>12</td>\n",
       "<td>0.5991285</td>\n",
       "<td>0.0976293</td>\n",
       "<td>0.0</td>\n",
       "<td>1.6690909</td>\n",
       "<td>0.0</td>\n",
       "<td>0.5672727</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>-100.0</td>\n",
       "<td>66.9090909</td></tr>\n",
       "<tr><td></td>\n",
       "<td>13</td>\n",
       "<td>0.6993464</td>\n",
       "<td>0.0697651</td>\n",
       "<td>0.0</td>\n",
       "<td>1.4299065</td>\n",
       "<td>0.0</td>\n",
       "<td>0.4859813</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>-100.0</td>\n",
       "<td>42.9906542</td></tr>\n",
       "<tr><td></td>\n",
       "<td>14</td>\n",
       "<td>0.7995643</td>\n",
       "<td>0.0454162</td>\n",
       "<td>0.0</td>\n",
       "<td>1.2506812</td>\n",
       "<td>0.0</td>\n",
       "<td>0.4250681</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>-100.0</td>\n",
       "<td>25.0681199</td></tr>\n",
       "<tr><td></td>\n",
       "<td>15</td>\n",
       "<td>0.8997821</td>\n",
       "<td>0.0294107</td>\n",
       "<td>0.0</td>\n",
       "<td>1.1113801</td>\n",
       "<td>0.0</td>\n",
       "<td>0.3777240</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>-100.0</td>\n",
       "<td>11.1380145</td></tr>\n",
       "<tr><td></td>\n",
       "<td>16</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0089469</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.3398693</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>-100.0</td>\n",
       "<td>0.0</td></tr></table></div>"
      ],
      "text/plain": [
       "    group    cumulative_data_fraction    lower_threshold    lift       cumulative_lift    response_rate    cumulative_response_rate    capture_rate    cumulative_capture_rate    gain      cumulative_gain\n",
       "--  -------  --------------------------  -----------------  ---------  -----------------  ---------------  --------------------------  --------------  -------------------------  --------  -----------------\n",
       "    1        0.0108932                   0.956728           2.94231    2.94231            1                1                           0.0320513       0.0320513                  194.231   194.231\n",
       "    2        0.0217865                   0.950477           2.94231    2.94231            1                1                           0.0320513       0.0641026                  194.231   194.231\n",
       "    3        0.0305011                   0.939277           2.94231    2.94231            1                1                           0.025641        0.0897436                  194.231   194.231\n",
       "    4        0.0413943                   0.928906           2.94231    2.94231            1                1                           0.0320513       0.121795                   194.231   194.231\n",
       "    5        0.0501089                   0.919977           2.94231    2.94231            1                1                           0.025641        0.147436                   194.231   194.231\n",
       "    6        0.100218                    0.890673           2.94231    2.94231            1                1                           0.147436        0.294872                   194.231   194.231\n",
       "    7        0.150327                    0.859338           2.94231    2.94231            1                1                           0.147436        0.442308                   194.231   194.231\n",
       "    8        0.200436                    0.761695           2.94231    2.94231            1                1                           0.147436        0.589744                   194.231   194.231\n",
       "    9        0.300654                    0.536646           2.87834    2.92099            0.978261         0.992754                    0.288462        0.878205                   187.834   192.099\n",
       "    10       0.400871                    0.29472            1.15134    2.47857            0.391304         0.842391                    0.115385        0.99359                    15.1338   147.857\n",
       "    11       0.501089                    0.177861           0.0639632  1.99565            0.0217391        0.678261                    0.00641026      1                          -93.6037  99.5652\n",
       "    12       0.599129                    0.0976293          0          1.66909            0                0.567273                    0               1                          -100      66.9091\n",
       "    13       0.699346                    0.0697651          0          1.42991            0                0.485981                    0               1                          -100      42.9907\n",
       "    14       0.799564                    0.0454162          0          1.25068            0                0.425068                    0               1                          -100      25.0681\n",
       "    15       0.899782                    0.0294107          0          1.11138            0                0.377724                    0               1                          -100      11.138\n",
       "    16       1                           0.00894693         0          1                  0                0.339869                    0               1                          -100      0"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Scoring History: \n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b></b></td>\n",
       "<td><b>timestamp</b></td>\n",
       "<td><b>duration</b></td>\n",
       "<td><b>number_of_trees</b></td>\n",
       "<td><b>training_rmse</b></td>\n",
       "<td><b>training_logloss</b></td>\n",
       "<td><b>training_auc</b></td>\n",
       "<td><b>training_lift</b></td>\n",
       "<td><b>training_classification_error</b></td></tr>\n",
       "<tr><td></td>\n",
       "<td>2018-08-13 17:56:05</td>\n",
       "<td> 0.022 sec</td>\n",
       "<td>0.0</td>\n",
       "<td>0.4736646</td>\n",
       "<td>0.6409487</td>\n",
       "<td>0.5</td>\n",
       "<td>1.0</td>\n",
       "<td>0.6601307</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2018-08-13 17:56:05</td>\n",
       "<td> 0.149 sec</td>\n",
       "<td>1.0</td>\n",
       "<td>0.4520203</td>\n",
       "<td>0.5968632</td>\n",
       "<td>0.8940721</td>\n",
       "<td>2.9423077</td>\n",
       "<td>0.1786492</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2018-08-13 17:56:05</td>\n",
       "<td> 0.174 sec</td>\n",
       "<td>2.0</td>\n",
       "<td>0.4335934</td>\n",
       "<td>0.5612398</td>\n",
       "<td>0.8995726</td>\n",
       "<td>2.9423077</td>\n",
       "<td>0.1851852</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2018-08-13 17:56:05</td>\n",
       "<td> 0.192 sec</td>\n",
       "<td>3.0</td>\n",
       "<td>0.4178224</td>\n",
       "<td>0.5316391</td>\n",
       "<td>0.9124566</td>\n",
       "<td>2.9423077</td>\n",
       "<td>0.1721133</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2018-08-13 17:56:05</td>\n",
       "<td> 0.203 sec</td>\n",
       "<td>4.0</td>\n",
       "<td>0.4032895</td>\n",
       "<td>0.5052655</td>\n",
       "<td>0.9252348</td>\n",
       "<td>2.9423077</td>\n",
       "<td>0.1372549</td></tr>\n",
       "<tr><td>---</td>\n",
       "<td>---</td>\n",
       "<td>---</td>\n",
       "<td>---</td>\n",
       "<td>---</td>\n",
       "<td>---</td>\n",
       "<td>---</td>\n",
       "<td>---</td>\n",
       "<td>---</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2018-08-13 17:56:05</td>\n",
       "<td> 0.601 sec</td>\n",
       "<td>46.0</td>\n",
       "<td>0.2215130</td>\n",
       "<td>0.2014016</td>\n",
       "<td>0.9946687</td>\n",
       "<td>2.9423077</td>\n",
       "<td>0.0326797</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2018-08-13 17:56:05</td>\n",
       "<td> 0.622 sec</td>\n",
       "<td>47.0</td>\n",
       "<td>0.2187999</td>\n",
       "<td>0.1980304</td>\n",
       "<td>0.9951553</td>\n",
       "<td>2.9423077</td>\n",
       "<td>0.0305011</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2018-08-13 17:56:05</td>\n",
       "<td> 0.628 sec</td>\n",
       "<td>48.0</td>\n",
       "<td>0.2167711</td>\n",
       "<td>0.1954565</td>\n",
       "<td>0.9955361</td>\n",
       "<td>2.9423077</td>\n",
       "<td>0.0305011</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2018-08-13 17:56:05</td>\n",
       "<td> 0.634 sec</td>\n",
       "<td>49.0</td>\n",
       "<td>0.2139505</td>\n",
       "<td>0.1915256</td>\n",
       "<td>0.9959592</td>\n",
       "<td>2.9423077</td>\n",
       "<td>0.0326797</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2018-08-13 17:56:05</td>\n",
       "<td> 0.647 sec</td>\n",
       "<td>50.0</td>\n",
       "<td>0.2129851</td>\n",
       "<td>0.1900386</td>\n",
       "<td>0.9959275</td>\n",
       "<td>2.9423077</td>\n",
       "<td>0.0305011</td></tr></table></div>"
      ],
      "text/plain": [
       "     timestamp            duration    number_of_trees    training_rmse        training_logloss     training_auc        training_lift      training_classification_error\n",
       "---  -------------------  ----------  -----------------  -------------------  -------------------  ------------------  -----------------  -------------------------------\n",
       "     2018-08-13 17:56:05  0.022 sec   0.0                0.47366459952924167  0.6409487346787026   0.5                 1.0                0.6601307189542484\n",
       "     2018-08-13 17:56:05  0.149 sec   1.0                0.45202026084797053  0.5968632425099655   0.894072099517644   2.942307692307692  0.1786492374727669\n",
       "     2018-08-13 17:56:05  0.174 sec   2.0                0.43359338111001383  0.5612397917100204   0.8995726495726496  2.942307692307692  0.18518518518518517\n",
       "     2018-08-13 17:56:05  0.192 sec   3.0                0.41782240631303796  0.5316390928737247   0.9124566302784124  2.942307692307692  0.1721132897603486\n",
       "     2018-08-13 17:56:05  0.203 sec   4.0                0.403289469859403    0.5052655452217871   0.9252348311754253  2.942307692307692  0.13725490196078433\n",
       "---  ---                  ---         ---                ---                  ---                  ---                 ---                ---\n",
       "     2018-08-13 17:56:05  0.601 sec   46.0               0.2215130217443874   0.20140157384660046  0.9946686976389946  2.942307692307692  0.032679738562091505\n",
       "     2018-08-13 17:56:05  0.622 sec   47.0               0.21879992219990438  0.19803037693180434  0.9951552847592451  2.942307692307692  0.030501089324618737\n",
       "     2018-08-13 17:56:05  0.628 sec   48.0               0.21677107206019575  0.1954564884035344   0.9955360920707456  2.942307692307692  0.030501089324618737\n",
       "     2018-08-13 17:56:05  0.634 sec   49.0               0.21395045933985735  0.19152559541891623  0.995959211305746   2.942307692307692  0.032679738562091505\n",
       "     2018-08-13 17:56:05  0.647 sec   50.0               0.21298514796088547  0.19003856764530652  0.9959274773631209  2.942307692307692  0.030501089324618737"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "See the whole table with table.as_data_frame()\n",
      "Variable Importances: \n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b>variable</b></td>\n",
       "<td><b>relative_importance</b></td>\n",
       "<td><b>scaled_importance</b></td>\n",
       "<td><b>percentage</b></td></tr>\n",
       "<tr><td>Glucose</td>\n",
       "<td>131.2686920</td>\n",
       "<td>1.0</td>\n",
       "<td>0.3196842</td></tr>\n",
       "<tr><td>BMI</td>\n",
       "<td>96.5606995</td>\n",
       "<td>0.7355958</td>\n",
       "<td>0.2351583</td></tr>\n",
       "<tr><td>DiabetesPedigreeFunction</td>\n",
       "<td>42.3217583</td>\n",
       "<td>0.3224056</td>\n",
       "<td>0.1030680</td></tr>\n",
       "<tr><td>Age</td>\n",
       "<td>36.2873573</td>\n",
       "<td>0.2764357</td>\n",
       "<td>0.0883721</td></tr>\n",
       "<tr><td>Pregnancies</td>\n",
       "<td>32.7109032</td>\n",
       "<td>0.2491904</td>\n",
       "<td>0.0796622</td></tr>\n",
       "<tr><td>BloodPressure</td>\n",
       "<td>27.1474304</td>\n",
       "<td>0.2068081</td>\n",
       "<td>0.0661133</td></tr>\n",
       "<tr><td>Insulin</td>\n",
       "<td>26.3434982</td>\n",
       "<td>0.2006838</td>\n",
       "<td>0.0641554</td></tr>\n",
       "<tr><td>SkinThickness</td>\n",
       "<td>17.9795818</td>\n",
       "<td>0.1369678</td>\n",
       "<td>0.0437864</td></tr></table></div>"
      ],
      "text/plain": [
       "variable                  relative_importance    scaled_importance    percentage\n",
       "------------------------  ---------------------  -------------------  ------------\n",
       "Glucose                   131.269                1                    0.319684\n",
       "BMI                       96.5607                0.735596             0.235158\n",
       "DiabetesPedigreeFunction  42.3218                0.322406             0.103068\n",
       "Age                       36.2874                0.276436             0.0883721\n",
       "Pregnancies               32.7109                0.24919              0.0796622\n",
       "BloodPressure             27.1474                0.206808             0.0661133\n",
       "Insulin                   26.3435                0.200684             0.0641554\n",
       "SkinThickness             17.9796                0.136968             0.0437864"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "print(gbm)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Time For Some Model Performance Evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "ModelMetricsBinomial: gbm\n",
      "** Reported on test data. **\n",
      "\n",
      "MSE: 0.15477851347832733\n",
      "RMSE: 0.393419004978569\n",
      "LogLoss: 0.4712061913524918\n",
      "Mean Per-Class Error: 0.19827956989247308\n",
      "AUC: 0.8408602150537635\n",
      "Gini: 0.681720430107527\n",
      "Confusion Matrix (Act/Pred) for max f1 @ threshold = 0.29535891405252496: \n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b></b></td>\n",
       "<td><b>0</b></td>\n",
       "<td><b>1</b></td>\n",
       "<td><b>Error</b></td>\n",
       "<td><b>Rate</b></td></tr>\n",
       "<tr><td>0</td>\n",
       "<td>71.0</td>\n",
       "<td>22.0</td>\n",
       "<td>0.2366</td>\n",
       "<td> (22.0/93.0)</td></tr>\n",
       "<tr><td>1</td>\n",
       "<td>8.0</td>\n",
       "<td>42.0</td>\n",
       "<td>0.16</td>\n",
       "<td> (8.0/50.0)</td></tr>\n",
       "<tr><td>Total</td>\n",
       "<td>79.0</td>\n",
       "<td>64.0</td>\n",
       "<td>0.2098</td>\n",
       "<td> (30.0/143.0)</td></tr></table></div>"
      ],
      "text/plain": [
       "       0    1    Error    Rate\n",
       "-----  ---  ---  -------  ------------\n",
       "0      71   22   0.2366   (22.0/93.0)\n",
       "1      8    42   0.16     (8.0/50.0)\n",
       "Total  79   64   0.2098   (30.0/143.0)"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Maximum Metrics: Maximum metrics at their respective thresholds\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b>metric</b></td>\n",
       "<td><b>threshold</b></td>\n",
       "<td><b>value</b></td>\n",
       "<td><b>idx</b></td></tr>\n",
       "<tr><td>max f1</td>\n",
       "<td>0.2953589</td>\n",
       "<td>0.7368421</td>\n",
       "<td>63.0</td></tr>\n",
       "<tr><td>max f2</td>\n",
       "<td>0.0569472</td>\n",
       "<td>0.8196721</td>\n",
       "<td>104.0</td></tr>\n",
       "<tr><td>max f0point5</td>\n",
       "<td>0.5779027</td>\n",
       "<td>0.7142857</td>\n",
       "<td>39.0</td></tr>\n",
       "<tr><td>max accuracy</td>\n",
       "<td>0.5779027</td>\n",
       "<td>0.7902098</td>\n",
       "<td>39.0</td></tr>\n",
       "<tr><td>max precision</td>\n",
       "<td>0.9649770</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td></tr>\n",
       "<tr><td>max recall</td>\n",
       "<td>0.0569472</td>\n",
       "<td>1.0</td>\n",
       "<td>104.0</td></tr>\n",
       "<tr><td>max specificity</td>\n",
       "<td>0.9649770</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td></tr>\n",
       "<tr><td>max absolute_mcc</td>\n",
       "<td>0.2953589</td>\n",
       "<td>0.5787056</td>\n",
       "<td>63.0</td></tr>\n",
       "<tr><td>max min_per_class_accuracy</td>\n",
       "<td>0.3421577</td>\n",
       "<td>0.7634409</td>\n",
       "<td>60.0</td></tr>\n",
       "<tr><td>max mean_per_class_accuracy</td>\n",
       "<td>0.2953589</td>\n",
       "<td>0.8017204</td>\n",
       "<td>63.0</td></tr></table></div>"
      ],
      "text/plain": [
       "metric                       threshold    value     idx\n",
       "---------------------------  -----------  --------  -----\n",
       "max f1                       0.295359     0.736842  63\n",
       "max f2                       0.0569472    0.819672  104\n",
       "max f0point5                 0.577903     0.714286  39\n",
       "max accuracy                 0.577903     0.79021   39\n",
       "max precision                0.964977     1         0\n",
       "max recall                   0.0569472    1         104\n",
       "max specificity              0.964977     1         0\n",
       "max absolute_mcc             0.295359     0.578706  63\n",
       "max min_per_class_accuracy   0.342158     0.763441  60\n",
       "max mean_per_class_accuracy  0.295359     0.80172   63"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Gains/Lift Table: Avg response rate: 34.97 %\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b></b></td>\n",
       "<td><b>group</b></td>\n",
       "<td><b>cumulative_data_fraction</b></td>\n",
       "<td><b>lower_threshold</b></td>\n",
       "<td><b>lift</b></td>\n",
       "<td><b>cumulative_lift</b></td>\n",
       "<td><b>response_rate</b></td>\n",
       "<td><b>cumulative_response_rate</b></td>\n",
       "<td><b>capture_rate</b></td>\n",
       "<td><b>cumulative_capture_rate</b></td>\n",
       "<td><b>gain</b></td>\n",
       "<td><b>cumulative_gain</b></td></tr>\n",
       "<tr><td></td>\n",
       "<td>1</td>\n",
       "<td>0.0139860</td>\n",
       "<td>0.9087377</td>\n",
       "<td>2.8600000</td>\n",
       "<td>2.8600000</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.04</td>\n",
       "<td>0.04</td>\n",
       "<td>186.0000000</td>\n",
       "<td>186.0000000</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2</td>\n",
       "<td>0.0209790</td>\n",
       "<td>0.9041231</td>\n",
       "<td>2.8600000</td>\n",
       "<td>2.8600000</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.02</td>\n",
       "<td>0.06</td>\n",
       "<td>186.0000000</td>\n",
       "<td>186.0000000</td></tr>\n",
       "<tr><td></td>\n",
       "<td>3</td>\n",
       "<td>0.0349650</td>\n",
       "<td>0.8895745</td>\n",
       "<td>1.4300000</td>\n",
       "<td>2.2880000</td>\n",
       "<td>0.5</td>\n",
       "<td>0.8</td>\n",
       "<td>0.02</td>\n",
       "<td>0.08</td>\n",
       "<td>43.0000000</td>\n",
       "<td>128.8</td></tr>\n",
       "<tr><td></td>\n",
       "<td>4</td>\n",
       "<td>0.0419580</td>\n",
       "<td>0.8863926</td>\n",
       "<td>0.0</td>\n",
       "<td>1.9066667</td>\n",
       "<td>0.0</td>\n",
       "<td>0.6666667</td>\n",
       "<td>0.0</td>\n",
       "<td>0.08</td>\n",
       "<td>-100.0</td>\n",
       "<td>90.6666667</td></tr>\n",
       "<tr><td></td>\n",
       "<td>5</td>\n",
       "<td>0.0559441</td>\n",
       "<td>0.8808375</td>\n",
       "<td>1.4300000</td>\n",
       "<td>1.7875</td>\n",
       "<td>0.5</td>\n",
       "<td>0.625</td>\n",
       "<td>0.02</td>\n",
       "<td>0.1</td>\n",
       "<td>43.0000000</td>\n",
       "<td>78.7500000</td></tr>\n",
       "<tr><td></td>\n",
       "<td>6</td>\n",
       "<td>0.1048951</td>\n",
       "<td>0.8323606</td>\n",
       "<td>2.8600000</td>\n",
       "<td>2.2880000</td>\n",
       "<td>1.0</td>\n",
       "<td>0.8</td>\n",
       "<td>0.14</td>\n",
       "<td>0.24</td>\n",
       "<td>186.0000000</td>\n",
       "<td>128.8</td></tr>\n",
       "<tr><td></td>\n",
       "<td>7</td>\n",
       "<td>0.1538462</td>\n",
       "<td>0.7336622</td>\n",
       "<td>1.6342857</td>\n",
       "<td>2.08</td>\n",
       "<td>0.5714286</td>\n",
       "<td>0.7272727</td>\n",
       "<td>0.08</td>\n",
       "<td>0.32</td>\n",
       "<td>63.4285714</td>\n",
       "<td>108.0</td></tr>\n",
       "<tr><td></td>\n",
       "<td>8</td>\n",
       "<td>0.2027972</td>\n",
       "<td>0.6898755</td>\n",
       "<td>1.6342857</td>\n",
       "<td>1.9724138</td>\n",
       "<td>0.5714286</td>\n",
       "<td>0.6896552</td>\n",
       "<td>0.08</td>\n",
       "<td>0.4</td>\n",
       "<td>63.4285714</td>\n",
       "<td>97.2413793</td></tr>\n",
       "<tr><td></td>\n",
       "<td>9</td>\n",
       "<td>0.3006993</td>\n",
       "<td>0.5466329</td>\n",
       "<td>2.2471429</td>\n",
       "<td>2.0618605</td>\n",
       "<td>0.7857143</td>\n",
       "<td>0.7209302</td>\n",
       "<td>0.22</td>\n",
       "<td>0.62</td>\n",
       "<td>124.7142857</td>\n",
       "<td>106.1860465</td></tr>\n",
       "<tr><td></td>\n",
       "<td>10</td>\n",
       "<td>0.3986014</td>\n",
       "<td>0.3672284</td>\n",
       "<td>1.4300000</td>\n",
       "<td>1.9066667</td>\n",
       "<td>0.5</td>\n",
       "<td>0.6666667</td>\n",
       "<td>0.14</td>\n",
       "<td>0.76</td>\n",
       "<td>43.0000000</td>\n",
       "<td>90.6666667</td></tr>\n",
       "<tr><td></td>\n",
       "<td>11</td>\n",
       "<td>0.5034965</td>\n",
       "<td>0.2278049</td>\n",
       "<td>0.7626667</td>\n",
       "<td>1.6683333</td>\n",
       "<td>0.2666667</td>\n",
       "<td>0.5833333</td>\n",
       "<td>0.08</td>\n",
       "<td>0.84</td>\n",
       "<td>-23.7333333</td>\n",
       "<td>66.8333333</td></tr>\n",
       "<tr><td></td>\n",
       "<td>12</td>\n",
       "<td>0.6013986</td>\n",
       "<td>0.1260237</td>\n",
       "<td>0.6128571</td>\n",
       "<td>1.4965116</td>\n",
       "<td>0.2142857</td>\n",
       "<td>0.5232558</td>\n",
       "<td>0.06</td>\n",
       "<td>0.9</td>\n",
       "<td>-38.7142857</td>\n",
       "<td>49.6511628</td></tr>\n",
       "<tr><td></td>\n",
       "<td>13</td>\n",
       "<td>0.6993007</td>\n",
       "<td>0.0774367</td>\n",
       "<td>0.4085714</td>\n",
       "<td>1.3442</td>\n",
       "<td>0.1428571</td>\n",
       "<td>0.47</td>\n",
       "<td>0.04</td>\n",
       "<td>0.94</td>\n",
       "<td>-59.1428571</td>\n",
       "<td>34.4200000</td></tr>\n",
       "<tr><td></td>\n",
       "<td>14</td>\n",
       "<td>0.7972028</td>\n",
       "<td>0.0424600</td>\n",
       "<td>0.6128571</td>\n",
       "<td>1.2543860</td>\n",
       "<td>0.2142857</td>\n",
       "<td>0.4385965</td>\n",
       "<td>0.06</td>\n",
       "<td>1.0</td>\n",
       "<td>-38.7142857</td>\n",
       "<td>25.4385965</td></tr>\n",
       "<tr><td></td>\n",
       "<td>15</td>\n",
       "<td>0.8951049</td>\n",
       "<td>0.0268626</td>\n",
       "<td>0.0</td>\n",
       "<td>1.1171875</td>\n",
       "<td>0.0</td>\n",
       "<td>0.390625</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>-100.0</td>\n",
       "<td>11.71875</td></tr>\n",
       "<tr><td></td>\n",
       "<td>16</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0144205</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.3496503</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>-100.0</td>\n",
       "<td>0.0</td></tr></table></div>"
      ],
      "text/plain": [
       "    group    cumulative_data_fraction    lower_threshold    lift      cumulative_lift    response_rate    cumulative_response_rate    capture_rate    cumulative_capture_rate    gain      cumulative_gain\n",
       "--  -------  --------------------------  -----------------  --------  -----------------  ---------------  --------------------------  --------------  -------------------------  --------  -----------------\n",
       "    1        0.013986                    0.908738           2.86      2.86               1                1                           0.04            0.04                       186       186\n",
       "    2        0.020979                    0.904123           2.86      2.86               1                1                           0.02            0.06                       186       186\n",
       "    3        0.034965                    0.889574           1.43      2.288              0.5              0.8                         0.02            0.08                       43        128.8\n",
       "    4        0.041958                    0.886393           0         1.90667            0                0.666667                    0               0.08                       -100      90.6667\n",
       "    5        0.0559441                   0.880838           1.43      1.7875             0.5              0.625                       0.02            0.1                        43        78.75\n",
       "    6        0.104895                    0.832361           2.86      2.288              1                0.8                         0.14            0.24                       186       128.8\n",
       "    7        0.153846                    0.733662           1.63429   2.08               0.571429         0.727273                    0.08            0.32                       63.4286   108\n",
       "    8        0.202797                    0.689876           1.63429   1.97241            0.571429         0.689655                    0.08            0.4                        63.4286   97.2414\n",
       "    9        0.300699                    0.546633           2.24714   2.06186            0.785714         0.72093                     0.22            0.62                       124.714   106.186\n",
       "    10       0.398601                    0.367228           1.43      1.90667            0.5              0.666667                    0.14            0.76                       43        90.6667\n",
       "    11       0.503497                    0.227805           0.762667  1.66833            0.266667         0.583333                    0.08            0.84                       -23.7333  66.8333\n",
       "    12       0.601399                    0.126024           0.612857  1.49651            0.214286         0.523256                    0.06            0.9                        -38.7143  49.6512\n",
       "    13       0.699301                    0.0774367          0.408571  1.3442             0.142857         0.47                        0.04            0.94                       -59.1429  34.42\n",
       "    14       0.797203                    0.04246            0.612857  1.25439            0.214286         0.438596                    0.06            1                          -38.7143  25.4386\n",
       "    15       0.895105                    0.0268626          0         1.11719            0                0.390625                    0               1                          -100      11.7188\n",
       "    16       1                           0.0144205          0         1                  0                0.34965                     0               1                          -100      0"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "evaluation = gbm.model_performance(test_data = val)\n",
    "print(evaluation)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "gbm Model Build progress: |███████████████████████████████████████████████| 100%\n"
     ]
    }
   ],
   "source": [
    "gbm_tune = H2OGradientBoostingEstimator(\n",
    "    ntrees = 3000,\n",
    "    learn_rate = 0.01,\n",
    "    stopping_rounds = 20,\n",
    "    stopping_metric = 'AUC',\n",
    "    col_sample_rate = 0.7,\n",
    "    seed = 100\n",
    ")\n",
    "\n",
    "gbm_tune.train(x = feature_variables, y = response, training_frame = train, validation_frame = val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8531182795698925"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gbm_tune.model_performance(val).auc()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Almost similiar performance after using the validation and early stoping as well.\n",
    "\n",
    "Now using `GridSearchCV`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "from h2o.grid.grid_search import H2OGridSearch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "gbm_grid = H2OGradientBoostingEstimator(\n",
    "    ntrees = 3000,\n",
    "    learn_rate = 0.01,\n",
    "    stopping_rounds = 20,\n",
    "    stopping_metric = 'AUC',\n",
    "    col_sample_rate = 0.7,\n",
    "    seed = 100\n",
    ")\n",
    "\n",
    "hyper_params = {'max_depth' : [4, 6, 8, 10, 12]}\n",
    "\n",
    "gbm_grid_search = H2OGridSearch(model = gbm_grid, \n",
    "                                hyper_params = hyper_params,\n",
    "                                grid_id = 'GRID_SEARCH_1',\n",
    "                                search_criteria = {'strategy' : 'Cartesian'}\n",
    "                               )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "gbm Grid Build progress: |████████████████████████████████████████████████| 100%\n"
     ]
    }
   ],
   "source": [
    "gbm_grid_search.train(x = feature_variables, \n",
    "                      y = response, \n",
    "                      training_frame = train,\n",
    "                      validation_frame = val\n",
    "                     )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "    max_depth              model_ids              logloss\n",
      "0           6  GRID_SEARCH_1_model_1  0.46681062193597067\n",
      "1          12  GRID_SEARCH_1_model_4   0.4799385784230339\n",
      "2          10  GRID_SEARCH_1_model_3  0.47993864805213987\n",
      "3           8  GRID_SEARCH_1_model_2   0.5228623502299519\n",
      "4           4  GRID_SEARCH_1_model_0   0.5574248488607398\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(gbm_grid_search)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "    max_depth              model_ids                 auc\n",
      "0           6  GRID_SEARCH_1_model_1  0.8668817204301076\n",
      "1          12  GRID_SEARCH_1_model_4  0.8619354838709677\n",
      "2          10  GRID_SEARCH_1_model_3  0.8619354838709677\n",
      "3           8  GRID_SEARCH_1_model_2  0.8572043010752688\n",
      "4           4  GRID_SEARCH_1_model_0  0.8540860215053764\n",
      "\n"
     ]
    }
   ],
   "source": [
    "sorted_grid_auc = gbm_grid_search.get_grid(sort_by = 'auc', decreasing = True)\n",
    "print(sorted_grid_auc)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's now do some k fold cross validation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "gbm_cv = H2OGradientBoostingEstimator(\n",
    "    ntrees = 3000,\n",
    "    learn_rate = 0.01,\n",
    "    stopping_rounds = 20,\n",
    "    stopping_metric = 'AUC',\n",
    "    col_sample_rate = 0.7,\n",
    "    seed = 100,\n",
    "    nfolds = 4\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "gbm Model Build 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></th>\n",
       "      <th>mean</th>\n",
       "      <th>sd</th>\n",
       "      <th>cv_1_valid</th>\n",
       "      <th>cv_2_valid</th>\n",
       "      <th>cv_3_valid</th>\n",
       "      <th>cv_4_valid</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>accuracy</td>\n",
       "      <td>0.7825497</td>\n",
       "      <td>0.012661983</td>\n",
       "      <td>0.8041958</td>\n",
       "      <td>0.76666665</td>\n",
       "      <td>0.79617834</td>\n",
       "      <td>0.7631579</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>auc</td>\n",
       "      <td>0.8295362</td>\n",
       "      <td>0.008918779</td>\n",
       "      <td>0.8116135</td>\n",
       "      <td>0.8466435</td>\n",
       "      <td>0.833271</td>\n",
       "      <td>0.8266167</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>err</td>\n",
       "      <td>0.21745032</td>\n",
       "      <td>0.012661983</td>\n",
       "      <td>0.1958042</td>\n",
       "      <td>0.23333333</td>\n",
       "      <td>0.20382166</td>\n",
       "      <td>0.23684211</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>err_count</td>\n",
       "      <td>32.75</td>\n",
       "      <td>2.2008522</td>\n",
       "      <td>28.0</td>\n",
       "      <td>35.0</td>\n",
       "      <td>32.0</td>\n",
       "      <td>36.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>f0point5</td>\n",
       "      <td>0.6754768</td>\n",
       "      <td>0.011119075</td>\n",
       "      <td>0.70212764</td>\n",
       "      <td>0.6666667</td>\n",
       "      <td>0.67114097</td>\n",
       "      <td>0.6619718</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>f1</td>\n",
       "      <td>0.7137344</td>\n",
       "      <td>0.005306089</td>\n",
       "      <td>0.70212764</td>\n",
       "      <td>0.7154471</td>\n",
       "      <td>0.71428573</td>\n",
       "      <td>0.72307694</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>f2</td>\n",
       "      <td>0.7585066</td>\n",
       "      <td>0.02458199</td>\n",
       "      <td>0.70212764</td>\n",
       "      <td>0.7719298</td>\n",
       "      <td>0.7633588</td>\n",
       "      <td>0.7966102</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>lift_top_group</td>\n",
       "      <td>1.8109505</td>\n",
       "      <td>0.3917452</td>\n",
       "      <td>1.5212766</td>\n",
       "      <td>1.3888888</td>\n",
       "      <td>1.57</td>\n",
       "      <td>2.7636364</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>logloss</td>\n",
       "      <td>0.5427928</td>\n",
       "      <td>0.008509446</td>\n",
       "      <td>0.5335831</td>\n",
       "      <td>0.53486794</td>\n",
       "      <td>0.53942543</td>\n",
       "      <td>0.5632946</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>max_per_class_error</td>\n",
       "      <td>0.26313907</td>\n",
       "      <td>0.025433745</td>\n",
       "      <td>0.29787233</td>\n",
       "      <td>0.26041666</td>\n",
       "      <td>0.20560747</td>\n",
       "      <td>0.28865978</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>mcc</td>\n",
       "      <td>0.55015486</td>\n",
       "      <td>0.008709082</td>\n",
       "      <td>0.5562943</td>\n",
       "      <td>0.5339336</td>\n",
       "      <td>0.5664904</td>\n",
       "      <td>0.5439011</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>mean_per_class_accuracy</td>\n",
       "      <td>0.78387135</td>\n",
       "      <td>0.00565355</td>\n",
       "      <td>0.77814716</td>\n",
       "      <td>0.7771991</td>\n",
       "      <td>0.79719627</td>\n",
       "      <td>0.78294283</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>mean_per_class_error</td>\n",
       "      <td>0.21612866</td>\n",
       "      <td>0.00565355</td>\n",
       "      <td>0.22185284</td>\n",
       "      <td>0.22280093</td>\n",
       "      <td>0.20280373</td>\n",
       "      <td>0.21705717</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>mse</td>\n",
       "      <td>0.18061103</td>\n",
       "      <td>0.0037304794</td>\n",
       "      <td>0.17670077</td>\n",
       "      <td>0.17791292</td>\n",
       "      <td>0.17813043</td>\n",
       "      <td>0.18969996</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>precision</td>\n",
       "      <td>0.6529092</td>\n",
       "      <td>0.020624775</td>\n",
       "      <td>0.70212764</td>\n",
       "      <td>0.6376812</td>\n",
       "      <td>0.6451613</td>\n",
       "      <td>0.62666667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>r2</td>\n",
       "      <td>0.19618861</td>\n",
       "      <td>0.014175629</td>\n",
       "      <td>0.19916798</td>\n",
       "      <td>0.22780849</td>\n",
       "      <td>0.17930146</td>\n",
       "      <td>0.17847651</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>recall</td>\n",
       "      <td>0.792872</td>\n",
       "      <td>0.0396397</td>\n",
       "      <td>0.70212764</td>\n",
       "      <td>0.8148148</td>\n",
       "      <td>0.8</td>\n",
       "      <td>0.8545455</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>rmse</td>\n",
       "      <td>0.42493895</td>\n",
       "      <td>0.0043542213</td>\n",
       "      <td>0.4203579</td>\n",
       "      <td>0.42179725</td>\n",
       "      <td>0.422055</td>\n",
       "      <td>0.4355456</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>specificity</td>\n",
       "      <td>0.7748707</td>\n",
       "      <td>0.038649503</td>\n",
       "      <td>0.8541667</td>\n",
       "      <td>0.7395833</td>\n",
       "      <td>0.7943925</td>\n",
       "      <td>0.7113402</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                   mean            sd  cv_1_valid  cv_2_valid  \\\n",
       "0                  accuracy   0.7825497   0.012661983   0.8041958  0.76666665   \n",
       "1                       auc   0.8295362   0.008918779   0.8116135   0.8466435   \n",
       "2                       err  0.21745032   0.012661983   0.1958042  0.23333333   \n",
       "3                 err_count       32.75     2.2008522        28.0        35.0   \n",
       "4                  f0point5   0.6754768   0.011119075  0.70212764   0.6666667   \n",
       "5                        f1   0.7137344   0.005306089  0.70212764   0.7154471   \n",
       "6                        f2   0.7585066    0.02458199  0.70212764   0.7719298   \n",
       "7            lift_top_group   1.8109505     0.3917452   1.5212766   1.3888888   \n",
       "8                   logloss   0.5427928   0.008509446   0.5335831  0.53486794   \n",
       "9       max_per_class_error  0.26313907   0.025433745  0.29787233  0.26041666   \n",
       "10                      mcc  0.55015486   0.008709082   0.5562943   0.5339336   \n",
       "11  mean_per_class_accuracy  0.78387135    0.00565355  0.77814716   0.7771991   \n",
       "12     mean_per_class_error  0.21612866    0.00565355  0.22185284  0.22280093   \n",
       "13                      mse  0.18061103  0.0037304794  0.17670077  0.17791292   \n",
       "14                precision   0.6529092   0.020624775  0.70212764   0.6376812   \n",
       "15                       r2  0.19618861   0.014175629  0.19916798  0.22780849   \n",
       "16                   recall    0.792872     0.0396397  0.70212764   0.8148148   \n",
       "17                     rmse  0.42493895  0.0043542213   0.4203579  0.42179725   \n",
       "18              specificity   0.7748707   0.038649503   0.8541667   0.7395833   \n",
       "\n",
       "    cv_3_valid  cv_4_valid  \n",
       "0   0.79617834   0.7631579  \n",
       "1     0.833271   0.8266167  \n",
       "2   0.20382166  0.23684211  \n",
       "3         32.0        36.0  \n",
       "4   0.67114097   0.6619718  \n",
       "5   0.71428573  0.72307694  \n",
       "6    0.7633588   0.7966102  \n",
       "7         1.57   2.7636364  \n",
       "8   0.53942543   0.5632946  \n",
       "9   0.20560747  0.28865978  \n",
       "10   0.5664904   0.5439011  \n",
       "11  0.79719627  0.78294283  \n",
       "12  0.20280373  0.21705717  \n",
       "13  0.17813043  0.18969996  \n",
       "14   0.6451613  0.62666667  \n",
       "15  0.17930146  0.17847651  \n",
       "16         0.8   0.8545455  \n",
       "17    0.422055   0.4355456  \n",
       "18   0.7943925   0.7113402  "
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gbm_cv.train(x = feature_variables, y = response, training_frame = train.rbind(val))\n",
    "\n",
    "gbm_cv_summary = gbm_cv.cross_validation_metrics_summary().as_data_frame()\n",
    "\n",
    "gbm_cv_summary"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Feature Importance**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fd0704df9e8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "gbm_cv.varimp_plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Let's Try some AutoML stuff**\n",
    "\n",
    "H2O’s `AutoML` can be used for `automating the machine learning workflow`, which includes `automatic training` and `tuning of many models` within a user-specified time-limit. Stacked Ensembles will be automatically trained on collections of individual models to produce highly predictive ensemble models which, in most cases, will be the top performing models in the AutoML Leaderboard.\n",
    "\n",
    "So let us use the `H2OAutoML` function to do automatic machine learning. We can specify the `max_models` parameter which indicates the number of individual (or \"base\") models, and does not include the two ensemble models that are trained at the end."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "from h2o.automl import H2OAutoML"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AutoML progress: |████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n"
     ]
    }
   ],
   "source": [
    "aml = H2OAutoML(max_models = 10, seed = 100)\n",
    "aml.train(x = feature_variables, y = response, training_frame = train, validation_frame = val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table>\n",
       "<thead>\n",
       "<tr><th>model_id                                             </th><th style=\"text-align: right;\">     auc</th><th style=\"text-align: right;\">  logloss</th></tr>\n",
       "</thead>\n",
       "<tbody>\n",
       "<tr><td>StackedEnsemble_AllModels_0_AutoML_20180813_181226   </td><td style=\"text-align: right;\">0.819212</td><td style=\"text-align: right;\"> 0.494024</td></tr>\n",
       "<tr><td>StackedEnsemble_BestOfFamily_0_AutoML_20180813_181226</td><td style=\"text-align: right;\">0.819212</td><td style=\"text-align: right;\"> 0.494024</td></tr>\n",
       "<tr><td>StackedEnsemble_AllModels_0_AutoML_20180813_181327   </td><td style=\"text-align: right;\">0.819191</td><td style=\"text-align: right;\"> 0.49397 </td></tr>\n",
       "<tr><td>GLM_grid_0_AutoML_20180813_181327_model_0            </td><td style=\"text-align: right;\">0.811955</td><td style=\"text-align: right;\"> 0.501127</td></tr>\n",
       "<tr><td>GLM_grid_0_AutoML_20180813_181226_model_0            </td><td style=\"text-align: right;\">0.811955</td><td style=\"text-align: right;\"> 0.501127</td></tr>\n",
       "<tr><td>GBM_grid_0_AutoML_20180813_181327_model_4            </td><td style=\"text-align: right;\">0.811786</td><td style=\"text-align: right;\"> 0.4995  </td></tr>\n",
       "<tr><td>GBM_grid_0_AutoML_20180813_181226_model_4            </td><td style=\"text-align: right;\">0.811786</td><td style=\"text-align: right;\"> 0.4995  </td></tr>\n",
       "<tr><td>GBM_grid_0_AutoML_20180813_181327_model_2            </td><td style=\"text-align: right;\">0.786505</td><td style=\"text-align: right;\"> 0.525103</td></tr>\n",
       "<tr><td>GBM_grid_0_AutoML_20180813_181226_model_2            </td><td style=\"text-align: right;\">0.786505</td><td style=\"text-align: right;\"> 0.525103</td></tr>\n",
       "<tr><td>GBM_grid_0_AutoML_20180813_181226_model_0            </td><td style=\"text-align: right;\">0.784664</td><td style=\"text-align: right;\"> 0.54684 </td></tr>\n",
       "</tbody>\n",
       "</table>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": []
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lb = aml.leaderboard\n",
    "lb"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model Details\n",
      "=============\n",
      "H2OStackedEnsembleEstimator :  Stacked Ensemble\n",
      "Model Key:  StackedEnsemble_AllModels_0_AutoML_20180813_181226\n",
      "No model summary for this model\n",
      "\n",
      "\n",
      "ModelMetricsBinomialGLM: stackedensemble\n",
      "** Reported on train data. **\n",
      "\n",
      "MSE: 0.1281241628057207\n",
      "RMSE: 0.35794435713630224\n",
      "LogLoss: 0.4055300308372596\n",
      "Null degrees of freedom: 458\n",
      "Residual degrees of freedom: 454\n",
      "Null deviance: 588.3909384350488\n",
      "Residual deviance: 372.2765683086042\n",
      "AIC: 382.2765683086042\n",
      "AUC: 0.8932681729711432\n",
      "Gini: 0.7865363459422865\n",
      "Confusion Matrix (Act/Pred) for max f1 @ threshold = 0.2790319122728225: \n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b></b></td>\n",
       "<td><b>0</b></td>\n",
       "<td><b>1</b></td>\n",
       "<td><b>Error</b></td>\n",
       "<td><b>Rate</b></td></tr>\n",
       "<tr><td>0</td>\n",
       "<td>236.0</td>\n",
       "<td>67.0</td>\n",
       "<td>0.2211</td>\n",
       "<td> (67.0/303.0)</td></tr>\n",
       "<tr><td>1</td>\n",
       "<td>20.0</td>\n",
       "<td>136.0</td>\n",
       "<td>0.1282</td>\n",
       "<td> (20.0/156.0)</td></tr>\n",
       "<tr><td>Total</td>\n",
       "<td>256.0</td>\n",
       "<td>203.0</td>\n",
       "<td>0.1895</td>\n",
       "<td> (87.0/459.0)</td></tr></table></div>"
      ],
      "text/plain": [
       "       0    1    Error    Rate\n",
       "-----  ---  ---  -------  ------------\n",
       "0      236  67   0.2211   (67.0/303.0)\n",
       "1      20   136  0.1282   (20.0/156.0)\n",
       "Total  256  203  0.1895   (87.0/459.0)"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Maximum Metrics: Maximum metrics at their respective thresholds\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b>metric</b></td>\n",
       "<td><b>threshold</b></td>\n",
       "<td><b>value</b></td>\n",
       "<td><b>idx</b></td></tr>\n",
       "<tr><td>max f1</td>\n",
       "<td>0.2790319</td>\n",
       "<td>0.7576602</td>\n",
       "<td>196.0</td></tr>\n",
       "<tr><td>max f2</td>\n",
       "<td>0.1888358</td>\n",
       "<td>0.8371041</td>\n",
       "<td>246.0</td></tr>\n",
       "<tr><td>max f0point5</td>\n",
       "<td>0.6630670</td>\n",
       "<td>0.7661290</td>\n",
       "<td>81.0</td></tr>\n",
       "<tr><td>max accuracy</td>\n",
       "<td>0.3591034</td>\n",
       "<td>0.8148148</td>\n",
       "<td>170.0</td></tr>\n",
       "<tr><td>max precision</td>\n",
       "<td>0.9150704</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td></tr>\n",
       "<tr><td>max recall</td>\n",
       "<td>0.1096978</td>\n",
       "<td>1.0</td>\n",
       "<td>322.0</td></tr>\n",
       "<tr><td>max specificity</td>\n",
       "<td>0.9150704</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td></tr>\n",
       "<tr><td>max absolute_mcc</td>\n",
       "<td>0.2790319</td>\n",
       "<td>0.6205521</td>\n",
       "<td>196.0</td></tr>\n",
       "<tr><td>max min_per_class_accuracy</td>\n",
       "<td>0.3324407</td>\n",
       "<td>0.8076923</td>\n",
       "<td>176.0</td></tr>\n",
       "<tr><td>max mean_per_class_accuracy</td>\n",
       "<td>0.2790319</td>\n",
       "<td>0.8253364</td>\n",
       "<td>196.0</td></tr></table></div>"
      ],
      "text/plain": [
       "metric                       threshold    value     idx\n",
       "---------------------------  -----------  --------  -----\n",
       "max f1                       0.279032     0.75766   196\n",
       "max f2                       0.188836     0.837104  246\n",
       "max f0point5                 0.663067     0.766129  81\n",
       "max accuracy                 0.359103     0.814815  170\n",
       "max precision                0.91507      1         0\n",
       "max recall                   0.109698     1         322\n",
       "max specificity              0.91507      1         0\n",
       "max absolute_mcc             0.279032     0.620552  196\n",
       "max min_per_class_accuracy   0.332441     0.807692  176\n",
       "max mean_per_class_accuracy  0.279032     0.825336  196"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Gains/Lift Table: Avg response rate: 33.99 %\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b></b></td>\n",
       "<td><b>group</b></td>\n",
       "<td><b>cumulative_data_fraction</b></td>\n",
       "<td><b>lower_threshold</b></td>\n",
       "<td><b>lift</b></td>\n",
       "<td><b>cumulative_lift</b></td>\n",
       "<td><b>response_rate</b></td>\n",
       "<td><b>cumulative_response_rate</b></td>\n",
       "<td><b>capture_rate</b></td>\n",
       "<td><b>cumulative_capture_rate</b></td>\n",
       "<td><b>gain</b></td>\n",
       "<td><b>cumulative_gain</b></td></tr>\n",
       "<tr><td></td>\n",
       "<td>1</td>\n",
       "<td>0.0108932</td>\n",
       "<td>0.8851348</td>\n",
       "<td>2.9423077</td>\n",
       "<td>2.9423077</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0320513</td>\n",
       "<td>0.0320513</td>\n",
       "<td>194.2307692</td>\n",
       "<td>194.2307692</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2</td>\n",
       "<td>0.0217865</td>\n",
       "<td>0.8731837</td>\n",
       "<td>2.9423077</td>\n",
       "<td>2.9423077</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0320513</td>\n",
       "<td>0.0641026</td>\n",
       "<td>194.2307692</td>\n",
       "<td>194.2307692</td></tr>\n",
       "<tr><td></td>\n",
       "<td>3</td>\n",
       "<td>0.0305011</td>\n",
       "<td>0.8596565</td>\n",
       "<td>2.9423077</td>\n",
       "<td>2.9423077</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0256410</td>\n",
       "<td>0.0897436</td>\n",
       "<td>194.2307692</td>\n",
       "<td>194.2307692</td></tr>\n",
       "<tr><td></td>\n",
       "<td>4</td>\n",
       "<td>0.0413943</td>\n",
       "<td>0.8547971</td>\n",
       "<td>2.9423077</td>\n",
       "<td>2.9423077</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0320513</td>\n",
       "<td>0.1217949</td>\n",
       "<td>194.2307692</td>\n",
       "<td>194.2307692</td></tr>\n",
       "<tr><td></td>\n",
       "<td>5</td>\n",
       "<td>0.0501089</td>\n",
       "<td>0.8446621</td>\n",
       "<td>2.2067308</td>\n",
       "<td>2.8143813</td>\n",
       "<td>0.75</td>\n",
       "<td>0.9565217</td>\n",
       "<td>0.0192308</td>\n",
       "<td>0.1410256</td>\n",
       "<td>120.6730769</td>\n",
       "<td>181.4381271</td></tr>\n",
       "<tr><td></td>\n",
       "<td>6</td>\n",
       "<td>0.1002179</td>\n",
       "<td>0.7868694</td>\n",
       "<td>2.6864548</td>\n",
       "<td>2.7504181</td>\n",
       "<td>0.9130435</td>\n",
       "<td>0.9347826</td>\n",
       "<td>0.1346154</td>\n",
       "<td>0.2756410</td>\n",
       "<td>168.6454849</td>\n",
       "<td>175.0418060</td></tr>\n",
       "<tr><td></td>\n",
       "<td>7</td>\n",
       "<td>0.1503268</td>\n",
       "<td>0.7348912</td>\n",
       "<td>2.5585284</td>\n",
       "<td>2.6864548</td>\n",
       "<td>0.8695652</td>\n",
       "<td>0.9130435</td>\n",
       "<td>0.1282051</td>\n",
       "<td>0.4038462</td>\n",
       "<td>155.8528428</td>\n",
       "<td>168.6454849</td></tr>\n",
       "<tr><td></td>\n",
       "<td>8</td>\n",
       "<td>0.2004357</td>\n",
       "<td>0.6196986</td>\n",
       "<td>1.9188963</td>\n",
       "<td>2.4945652</td>\n",
       "<td>0.6521739</td>\n",
       "<td>0.8478261</td>\n",
       "<td>0.0961538</td>\n",
       "<td>0.5</td>\n",
       "<td>91.8896321</td>\n",
       "<td>149.4565217</td></tr>\n",
       "<tr><td></td>\n",
       "<td>9</td>\n",
       "<td>0.3006536</td>\n",
       "<td>0.4838111</td>\n",
       "<td>1.5351171</td>\n",
       "<td>2.1747492</td>\n",
       "<td>0.5217391</td>\n",
       "<td>0.7391304</td>\n",
       "<td>0.1538462</td>\n",
       "<td>0.6538462</td>\n",
       "<td>53.5117057</td>\n",
       "<td>117.4749164</td></tr>\n",
       "<tr><td></td>\n",
       "<td>10</td>\n",
       "<td>0.4008715</td>\n",
       "<td>0.3169329</td>\n",
       "<td>1.5351171</td>\n",
       "<td>2.0148411</td>\n",
       "<td>0.5217391</td>\n",
       "<td>0.6847826</td>\n",
       "<td>0.1538462</td>\n",
       "<td>0.8076923</td>\n",
       "<td>53.5117057</td>\n",
       "<td>101.4841137</td></tr>\n",
       "<tr><td></td>\n",
       "<td>11</td>\n",
       "<td>0.5010893</td>\n",
       "<td>0.2232572</td>\n",
       "<td>0.9594482</td>\n",
       "<td>1.8037625</td>\n",
       "<td>0.3260870</td>\n",
       "<td>0.6130435</td>\n",
       "<td>0.0961538</td>\n",
       "<td>0.9038462</td>\n",
       "<td>-4.0551839</td>\n",
       "<td>80.3762542</td></tr>\n",
       "<tr><td></td>\n",
       "<td>12</td>\n",
       "<td>0.5991285</td>\n",
       "<td>0.1726207</td>\n",
       "<td>0.5230769</td>\n",
       "<td>1.5941958</td>\n",
       "<td>0.1777778</td>\n",
       "<td>0.5418182</td>\n",
       "<td>0.0512821</td>\n",
       "<td>0.9551282</td>\n",
       "<td>-47.6923077</td>\n",
       "<td>59.4195804</td></tr>\n",
       "<tr><td></td>\n",
       "<td>13</td>\n",
       "<td>0.6993464</td>\n",
       "<td>0.1298552</td>\n",
       "<td>0.1918896</td>\n",
       "<td>1.3932423</td>\n",
       "<td>0.0652174</td>\n",
       "<td>0.4735202</td>\n",
       "<td>0.0192308</td>\n",
       "<td>0.9743590</td>\n",
       "<td>-80.8110368</td>\n",
       "<td>39.3242272</td></tr>\n",
       "<tr><td></td>\n",
       "<td>14</td>\n",
       "<td>0.7995643</td>\n",
       "<td>0.1040957</td>\n",
       "<td>0.2558528</td>\n",
       "<td>1.2506812</td>\n",
       "<td>0.0869565</td>\n",
       "<td>0.4250681</td>\n",
       "<td>0.0256410</td>\n",
       "<td>1.0</td>\n",
       "<td>-74.4147157</td>\n",
       "<td>25.0681199</td></tr>\n",
       "<tr><td></td>\n",
       "<td>15</td>\n",
       "<td>0.8997821</td>\n",
       "<td>0.0819700</td>\n",
       "<td>0.0</td>\n",
       "<td>1.1113801</td>\n",
       "<td>0.0</td>\n",
       "<td>0.3777240</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>-100.0</td>\n",
       "<td>11.1380145</td></tr>\n",
       "<tr><td></td>\n",
       "<td>16</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0666201</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.3398693</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>-100.0</td>\n",
       "<td>0.0</td></tr></table></div>"
      ],
      "text/plain": [
       "    group    cumulative_data_fraction    lower_threshold    lift      cumulative_lift    response_rate    cumulative_response_rate    capture_rate    cumulative_capture_rate    gain      cumulative_gain\n",
       "--  -------  --------------------------  -----------------  --------  -----------------  ---------------  --------------------------  --------------  -------------------------  --------  -----------------\n",
       "    1        0.0108932                   0.885135           2.94231   2.94231            1                1                           0.0320513       0.0320513                  194.231   194.231\n",
       "    2        0.0217865                   0.873184           2.94231   2.94231            1                1                           0.0320513       0.0641026                  194.231   194.231\n",
       "    3        0.0305011                   0.859656           2.94231   2.94231            1                1                           0.025641        0.0897436                  194.231   194.231\n",
       "    4        0.0413943                   0.854797           2.94231   2.94231            1                1                           0.0320513       0.121795                   194.231   194.231\n",
       "    5        0.0501089                   0.844662           2.20673   2.81438            0.75             0.956522                    0.0192308       0.141026                   120.673   181.438\n",
       "    6        0.100218                    0.786869           2.68645   2.75042            0.913043         0.934783                    0.134615        0.275641                   168.645   175.042\n",
       "    7        0.150327                    0.734891           2.55853   2.68645            0.869565         0.913043                    0.128205        0.403846                   155.853   168.645\n",
       "    8        0.200436                    0.619699           1.9189    2.49457            0.652174         0.847826                    0.0961538       0.5                        91.8896   149.457\n",
       "    9        0.300654                    0.483811           1.53512   2.17475            0.521739         0.73913                     0.153846        0.653846                   53.5117   117.475\n",
       "    10       0.400871                    0.316933           1.53512   2.01484            0.521739         0.684783                    0.153846        0.807692                   53.5117   101.484\n",
       "    11       0.501089                    0.223257           0.959448  1.80376            0.326087         0.613043                    0.0961538       0.903846                   -4.05518  80.3763\n",
       "    12       0.599129                    0.172621           0.523077  1.5942             0.177778         0.541818                    0.0512821       0.955128                   -47.6923  59.4196\n",
       "    13       0.699346                    0.129855           0.19189   1.39324            0.0652174        0.47352                     0.0192308       0.974359                   -80.811   39.3242\n",
       "    14       0.799564                    0.104096           0.255853  1.25068            0.0869565        0.425068                    0.025641        1                          -74.4147  25.0681\n",
       "    15       0.899782                    0.08197            0         1.11138            0                0.377724                    0               1                          -100      11.138\n",
       "    16       1                           0.0666201          0         1                  0                0.339869                    0               1                          -100      0"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "ModelMetricsBinomialGLM: stackedensemble\n",
      "** Reported on validation data. **\n",
      "\n",
      "MSE: 0.15816240073260845\n",
      "RMSE: 0.3976963675124635\n",
      "LogLoss: 0.4754155693116159\n",
      "Null degrees of freedom: 142\n",
      "Residual degrees of freedom: 138\n",
      "Null deviance: 185.16845752761944\n",
      "Residual deviance: 135.96885282312215\n",
      "AIC: 145.96885282312215\n",
      "AUC: 0.8473118279569892\n",
      "Gini: 0.6946236559139785\n",
      "Confusion Matrix (Act/Pred) for max f1 @ threshold = 0.21694837129030067: \n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b></b></td>\n",
       "<td><b>0</b></td>\n",
       "<td><b>1</b></td>\n",
       "<td><b>Error</b></td>\n",
       "<td><b>Rate</b></td></tr>\n",
       "<tr><td>0</td>\n",
       "<td>66.0</td>\n",
       "<td>27.0</td>\n",
       "<td>0.2903</td>\n",
       "<td> (27.0/93.0)</td></tr>\n",
       "<tr><td>1</td>\n",
       "<td>6.0</td>\n",
       "<td>44.0</td>\n",
       "<td>0.12</td>\n",
       "<td> (6.0/50.0)</td></tr>\n",
       "<tr><td>Total</td>\n",
       "<td>72.0</td>\n",
       "<td>71.0</td>\n",
       "<td>0.2308</td>\n",
       "<td> (33.0/143.0)</td></tr></table></div>"
      ],
      "text/plain": [
       "       0    1    Error    Rate\n",
       "-----  ---  ---  -------  ------------\n",
       "0      66   27   0.2903   (27.0/93.0)\n",
       "1      6    44   0.12     (6.0/50.0)\n",
       "Total  72   71   0.2308   (33.0/143.0)"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Maximum Metrics: Maximum metrics at their respective thresholds\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b>metric</b></td>\n",
       "<td><b>threshold</b></td>\n",
       "<td><b>value</b></td>\n",
       "<td><b>idx</b></td></tr>\n",
       "<tr><td>max f1</td>\n",
       "<td>0.2169484</td>\n",
       "<td>0.7272727</td>\n",
       "<td>70.0</td></tr>\n",
       "<tr><td>max f2</td>\n",
       "<td>0.1398948</td>\n",
       "<td>0.8390411</td>\n",
       "<td>91.0</td></tr>\n",
       "<tr><td>max f0point5</td>\n",
       "<td>0.6602723</td>\n",
       "<td>0.6962025</td>\n",
       "<td>26.0</td></tr>\n",
       "<tr><td>max accuracy</td>\n",
       "<td>0.6602723</td>\n",
       "<td>0.7692308</td>\n",
       "<td>26.0</td></tr>\n",
       "<tr><td>max precision</td>\n",
       "<td>0.8826838</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td></tr>\n",
       "<tr><td>max recall</td>\n",
       "<td>0.1263710</td>\n",
       "<td>1.0</td>\n",
       "<td>99.0</td></tr>\n",
       "<tr><td>max specificity</td>\n",
       "<td>0.8826838</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td></tr>\n",
       "<tr><td>max absolute_mcc</td>\n",
       "<td>0.2169484</td>\n",
       "<td>0.5624003</td>\n",
       "<td>70.0</td></tr>\n",
       "<tr><td>max min_per_class_accuracy</td>\n",
       "<td>0.2788580</td>\n",
       "<td>0.7526882</td>\n",
       "<td>60.0</td></tr>\n",
       "<tr><td>max mean_per_class_accuracy</td>\n",
       "<td>0.2169484</td>\n",
       "<td>0.7948387</td>\n",
       "<td>70.0</td></tr></table></div>"
      ],
      "text/plain": [
       "metric                       threshold    value     idx\n",
       "---------------------------  -----------  --------  -----\n",
       "max f1                       0.216948     0.727273  70\n",
       "max f2                       0.139895     0.839041  91\n",
       "max f0point5                 0.660272     0.696203  26\n",
       "max accuracy                 0.660272     0.769231  26\n",
       "max precision                0.882684     1         0\n",
       "max recall                   0.126371     1         99\n",
       "max specificity              0.882684     1         0\n",
       "max absolute_mcc             0.216948     0.5624    70\n",
       "max min_per_class_accuracy   0.278858     0.752688  60\n",
       "max mean_per_class_accuracy  0.216948     0.794839  70"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Gains/Lift Table: Avg response rate: 34.97 %\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b></b></td>\n",
       "<td><b>group</b></td>\n",
       "<td><b>cumulative_data_fraction</b></td>\n",
       "<td><b>lower_threshold</b></td>\n",
       "<td><b>lift</b></td>\n",
       "<td><b>cumulative_lift</b></td>\n",
       "<td><b>response_rate</b></td>\n",
       "<td><b>cumulative_response_rate</b></td>\n",
       "<td><b>capture_rate</b></td>\n",
       "<td><b>cumulative_capture_rate</b></td>\n",
       "<td><b>gain</b></td>\n",
       "<td><b>cumulative_gain</b></td></tr>\n",
       "<tr><td></td>\n",
       "<td>1</td>\n",
       "<td>0.0139860</td>\n",
       "<td>0.8654002</td>\n",
       "<td>2.8600000</td>\n",
       "<td>2.8600000</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.04</td>\n",
       "<td>0.04</td>\n",
       "<td>186.0000000</td>\n",
       "<td>186.0000000</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2</td>\n",
       "<td>0.0209790</td>\n",
       "<td>0.8397535</td>\n",
       "<td>0.0</td>\n",
       "<td>1.9066667</td>\n",
       "<td>0.0</td>\n",
       "<td>0.6666667</td>\n",
       "<td>0.0</td>\n",
       "<td>0.04</td>\n",
       "<td>-100.0</td>\n",
       "<td>90.6666667</td></tr>\n",
       "<tr><td></td>\n",
       "<td>3</td>\n",
       "<td>0.0349650</td>\n",
       "<td>0.8265332</td>\n",
       "<td>2.8600000</td>\n",
       "<td>2.2880000</td>\n",
       "<td>1.0</td>\n",
       "<td>0.8</td>\n",
       "<td>0.04</td>\n",
       "<td>0.08</td>\n",
       "<td>186.0000000</td>\n",
       "<td>128.8</td></tr>\n",
       "<tr><td></td>\n",
       "<td>4</td>\n",
       "<td>0.0419580</td>\n",
       "<td>0.8127614</td>\n",
       "<td>2.8600000</td>\n",
       "<td>2.3833333</td>\n",
       "<td>1.0</td>\n",
       "<td>0.8333333</td>\n",
       "<td>0.02</td>\n",
       "<td>0.1</td>\n",
       "<td>186.0000000</td>\n",
       "<td>138.3333333</td></tr>\n",
       "<tr><td></td>\n",
       "<td>5</td>\n",
       "<td>0.0559441</td>\n",
       "<td>0.8069624</td>\n",
       "<td>2.8600000</td>\n",
       "<td>2.5025</td>\n",
       "<td>1.0</td>\n",
       "<td>0.875</td>\n",
       "<td>0.04</td>\n",
       "<td>0.14</td>\n",
       "<td>186.0000000</td>\n",
       "<td>150.25</td></tr>\n",
       "<tr><td></td>\n",
       "<td>6</td>\n",
       "<td>0.1048951</td>\n",
       "<td>0.7683849</td>\n",
       "<td>1.6342857</td>\n",
       "<td>2.0973333</td>\n",
       "<td>0.5714286</td>\n",
       "<td>0.7333333</td>\n",
       "<td>0.08</td>\n",
       "<td>0.22</td>\n",
       "<td>63.4285714</td>\n",
       "<td>109.7333333</td></tr>\n",
       "<tr><td></td>\n",
       "<td>7</td>\n",
       "<td>0.1538462</td>\n",
       "<td>0.7061489</td>\n",
       "<td>2.4514286</td>\n",
       "<td>2.21</td>\n",
       "<td>0.8571429</td>\n",
       "<td>0.7727273</td>\n",
       "<td>0.12</td>\n",
       "<td>0.34</td>\n",
       "<td>145.1428571</td>\n",
       "<td>121.0</td></tr>\n",
       "<tr><td></td>\n",
       "<td>8</td>\n",
       "<td>0.2027972</td>\n",
       "<td>0.6362214</td>\n",
       "<td>2.4514286</td>\n",
       "<td>2.2682759</td>\n",
       "<td>0.8571429</td>\n",
       "<td>0.7931034</td>\n",
       "<td>0.12</td>\n",
       "<td>0.46</td>\n",
       "<td>145.1428571</td>\n",
       "<td>126.8275862</td></tr>\n",
       "<tr><td></td>\n",
       "<td>9</td>\n",
       "<td>0.3006993</td>\n",
       "<td>0.4377878</td>\n",
       "<td>1.2257143</td>\n",
       "<td>1.9288372</td>\n",
       "<td>0.4285714</td>\n",
       "<td>0.6744186</td>\n",
       "<td>0.12</td>\n",
       "<td>0.58</td>\n",
       "<td>22.5714286</td>\n",
       "<td>92.8837209</td></tr>\n",
       "<tr><td></td>\n",
       "<td>10</td>\n",
       "<td>0.3986014</td>\n",
       "<td>0.3275677</td>\n",
       "<td>1.2257143</td>\n",
       "<td>1.7561404</td>\n",
       "<td>0.4285714</td>\n",
       "<td>0.6140351</td>\n",
       "<td>0.12</td>\n",
       "<td>0.7</td>\n",
       "<td>22.5714286</td>\n",
       "<td>75.6140351</td></tr>\n",
       "<tr><td></td>\n",
       "<td>11</td>\n",
       "<td>0.5034965</td>\n",
       "<td>0.2020916</td>\n",
       "<td>1.716</td>\n",
       "<td>1.7477778</td>\n",
       "<td>0.6</td>\n",
       "<td>0.6111111</td>\n",
       "<td>0.18</td>\n",
       "<td>0.88</td>\n",
       "<td>71.6</td>\n",
       "<td>74.7777778</td></tr>\n",
       "<tr><td></td>\n",
       "<td>12</td>\n",
       "<td>0.6013986</td>\n",
       "<td>0.1594442</td>\n",
       "<td>0.6128571</td>\n",
       "<td>1.5630233</td>\n",
       "<td>0.2142857</td>\n",
       "<td>0.5465116</td>\n",
       "<td>0.06</td>\n",
       "<td>0.94</td>\n",
       "<td>-38.7142857</td>\n",
       "<td>56.3023256</td></tr>\n",
       "<tr><td></td>\n",
       "<td>13</td>\n",
       "<td>0.6993007</td>\n",
       "<td>0.1262696</td>\n",
       "<td>0.6128571</td>\n",
       "<td>1.4300000</td>\n",
       "<td>0.2142857</td>\n",
       "<td>0.5</td>\n",
       "<td>0.06</td>\n",
       "<td>1.0</td>\n",
       "<td>-38.7142857</td>\n",
       "<td>43.0000000</td></tr>\n",
       "<tr><td></td>\n",
       "<td>14</td>\n",
       "<td>0.7972028</td>\n",
       "<td>0.0923643</td>\n",
       "<td>0.0</td>\n",
       "<td>1.2543860</td>\n",
       "<td>0.0</td>\n",
       "<td>0.4385965</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>-100.0</td>\n",
       "<td>25.4385965</td></tr>\n",
       "<tr><td></td>\n",
       "<td>15</td>\n",
       "<td>0.8951049</td>\n",
       "<td>0.0771671</td>\n",
       "<td>0.0</td>\n",
       "<td>1.1171875</td>\n",
       "<td>0.0</td>\n",
       "<td>0.390625</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>-100.0</td>\n",
       "<td>11.71875</td></tr>\n",
       "<tr><td></td>\n",
       "<td>16</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0664558</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.3496503</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>-100.0</td>\n",
       "<td>0.0</td></tr></table></div>"
      ],
      "text/plain": [
       "    group    cumulative_data_fraction    lower_threshold    lift      cumulative_lift    response_rate    cumulative_response_rate    capture_rate    cumulative_capture_rate    gain      cumulative_gain\n",
       "--  -------  --------------------------  -----------------  --------  -----------------  ---------------  --------------------------  --------------  -------------------------  --------  -----------------\n",
       "    1        0.013986                    0.8654             2.86      2.86               1                1                           0.04            0.04                       186       186\n",
       "    2        0.020979                    0.839754           0         1.90667            0                0.666667                    0               0.04                       -100      90.6667\n",
       "    3        0.034965                    0.826533           2.86      2.288              1                0.8                         0.04            0.08                       186       128.8\n",
       "    4        0.041958                    0.812761           2.86      2.38333            1                0.833333                    0.02            0.1                        186       138.333\n",
       "    5        0.0559441                   0.806962           2.86      2.5025             1                0.875                       0.04            0.14                       186       150.25\n",
       "    6        0.104895                    0.768385           1.63429   2.09733            0.571429         0.733333                    0.08            0.22                       63.4286   109.733\n",
       "    7        0.153846                    0.706149           2.45143   2.21               0.857143         0.772727                    0.12            0.34                       145.143   121\n",
       "    8        0.202797                    0.636221           2.45143   2.26828            0.857143         0.793103                    0.12            0.46                       145.143   126.828\n",
       "    9        0.300699                    0.437788           1.22571   1.92884            0.428571         0.674419                    0.12            0.58                       22.5714   92.8837\n",
       "    10       0.398601                    0.327568           1.22571   1.75614            0.428571         0.614035                    0.12            0.7                        22.5714   75.614\n",
       "    11       0.503497                    0.202092           1.716     1.74778            0.6              0.611111                    0.18            0.88                       71.6      74.7778\n",
       "    12       0.601399                    0.159444           0.612857  1.56302            0.214286         0.546512                    0.06            0.94                       -38.7143  56.3023\n",
       "    13       0.699301                    0.12627            0.612857  1.43               0.214286         0.5                         0.06            1                          -38.7143  43\n",
       "    14       0.797203                    0.0923643          0         1.25439            0                0.438596                    0               1                          -100      25.4386\n",
       "    15       0.895105                    0.0771671          0         1.11719            0                0.390625                    0               1                          -100      11.7188\n",
       "    16       1                           0.0664558          0         1                  0                0.34965                     0               1                          -100      0"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "ModelMetricsBinomialGLM: stackedensemble\n",
      "** Reported on cross-validation data. **\n",
      "\n",
      "MSE: 0.16307639380264152\n",
      "RMSE: 0.40382718309029364\n",
      "LogLoss: 0.49402436452422505\n",
      "Null degrees of freedom: 458\n",
      "Residual degrees of freedom: 454\n",
      "Null deviance: 588.6009937909473\n",
      "Residual deviance: 453.5143666332385\n",
      "AIC: 463.5143666332385\n",
      "AUC: 0.8192117288651942\n",
      "Gini: 0.6384234577303884\n",
      "Confusion Matrix (Act/Pred) for max f1 @ threshold = 0.2467748082216652: \n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b></b></td>\n",
       "<td><b>0</b></td>\n",
       "<td><b>1</b></td>\n",
       "<td><b>Error</b></td>\n",
       "<td><b>Rate</b></td></tr>\n",
       "<tr><td>0</td>\n",
       "<td>209.0</td>\n",
       "<td>94.0</td>\n",
       "<td>0.3102</td>\n",
       "<td> (94.0/303.0)</td></tr>\n",
       "<tr><td>1</td>\n",
       "<td>27.0</td>\n",
       "<td>129.0</td>\n",
       "<td>0.1731</td>\n",
       "<td> (27.0/156.0)</td></tr>\n",
       "<tr><td>Total</td>\n",
       "<td>236.0</td>\n",
       "<td>223.0</td>\n",
       "<td>0.2636</td>\n",
       "<td> (121.0/459.0)</td></tr></table></div>"
      ],
      "text/plain": [
       "       0    1    Error    Rate\n",
       "-----  ---  ---  -------  -------------\n",
       "0      209  94   0.3102   (94.0/303.0)\n",
       "1      27   129  0.1731   (27.0/156.0)\n",
       "Total  236  223  0.2636   (121.0/459.0)"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Maximum Metrics: Maximum metrics at their respective thresholds\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b>metric</b></td>\n",
       "<td><b>threshold</b></td>\n",
       "<td><b>value</b></td>\n",
       "<td><b>idx</b></td></tr>\n",
       "<tr><td>max f1</td>\n",
       "<td>0.2467748</td>\n",
       "<td>0.6807388</td>\n",
       "<td>207.0</td></tr>\n",
       "<tr><td>max f2</td>\n",
       "<td>0.1640673</td>\n",
       "<td>0.7868852</td>\n",
       "<td>263.0</td></tr>\n",
       "<tr><td>max f0point5</td>\n",
       "<td>0.6329106</td>\n",
       "<td>0.6884058</td>\n",
       "<td>93.0</td></tr>\n",
       "<tr><td>max accuracy</td>\n",
       "<td>0.6329106</td>\n",
       "<td>0.7755991</td>\n",
       "<td>93.0</td></tr>\n",
       "<tr><td>max precision</td>\n",
       "<td>0.8967030</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td></tr>\n",
       "<tr><td>max recall</td>\n",
       "<td>0.0794502</td>\n",
       "<td>1.0</td>\n",
       "<td>376.0</td></tr>\n",
       "<tr><td>max specificity</td>\n",
       "<td>0.8967030</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td></tr>\n",
       "<tr><td>max absolute_mcc</td>\n",
       "<td>0.3261702</td>\n",
       "<td>0.4923973</td>\n",
       "<td>171.0</td></tr>\n",
       "<tr><td>max min_per_class_accuracy</td>\n",
       "<td>0.3052172</td>\n",
       "<td>0.75</td>\n",
       "<td>178.0</td></tr>\n",
       "<tr><td>max mean_per_class_accuracy</td>\n",
       "<td>0.2467748</td>\n",
       "<td>0.7583460</td>\n",
       "<td>207.0</td></tr></table></div>"
      ],
      "text/plain": [
       "metric                       threshold    value     idx\n",
       "---------------------------  -----------  --------  -----\n",
       "max f1                       0.246775     0.680739  207\n",
       "max f2                       0.164067     0.786885  263\n",
       "max f0point5                 0.632911     0.688406  93\n",
       "max accuracy                 0.632911     0.775599  93\n",
       "max precision                0.896703     1         0\n",
       "max recall                   0.0794502    1         376\n",
       "max specificity              0.896703     1         0\n",
       "max absolute_mcc             0.32617      0.492397  171\n",
       "max min_per_class_accuracy   0.305217     0.75      178\n",
       "max mean_per_class_accuracy  0.246775     0.758346  207"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Gains/Lift Table: Avg response rate: 33.99 %\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b></b></td>\n",
       "<td><b>group</b></td>\n",
       "<td><b>cumulative_data_fraction</b></td>\n",
       "<td><b>lower_threshold</b></td>\n",
       "<td><b>lift</b></td>\n",
       "<td><b>cumulative_lift</b></td>\n",
       "<td><b>response_rate</b></td>\n",
       "<td><b>cumulative_response_rate</b></td>\n",
       "<td><b>capture_rate</b></td>\n",
       "<td><b>cumulative_capture_rate</b></td>\n",
       "<td><b>gain</b></td>\n",
       "<td><b>cumulative_gain</b></td></tr>\n",
       "<tr><td></td>\n",
       "<td>1</td>\n",
       "<td>0.0108932</td>\n",
       "<td>0.8790950</td>\n",
       "<td>2.3538462</td>\n",
       "<td>2.3538462</td>\n",
       "<td>0.8</td>\n",
       "<td>0.8</td>\n",
       "<td>0.0256410</td>\n",
       "<td>0.0256410</td>\n",
       "<td>135.3846154</td>\n",
       "<td>135.3846154</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2</td>\n",
       "<td>0.0217865</td>\n",
       "<td>0.8564306</td>\n",
       "<td>2.3538462</td>\n",
       "<td>2.3538462</td>\n",
       "<td>0.8</td>\n",
       "<td>0.8</td>\n",
       "<td>0.0256410</td>\n",
       "<td>0.0512821</td>\n",
       "<td>135.3846154</td>\n",
       "<td>135.3846154</td></tr>\n",
       "<tr><td></td>\n",
       "<td>3</td>\n",
       "<td>0.0305011</td>\n",
       "<td>0.8333031</td>\n",
       "<td>2.2067308</td>\n",
       "<td>2.3118132</td>\n",
       "<td>0.75</td>\n",
       "<td>0.7857143</td>\n",
       "<td>0.0192308</td>\n",
       "<td>0.0705128</td>\n",
       "<td>120.6730769</td>\n",
       "<td>131.1813187</td></tr>\n",
       "<tr><td></td>\n",
       "<td>4</td>\n",
       "<td>0.0413943</td>\n",
       "<td>0.8293527</td>\n",
       "<td>2.3538462</td>\n",
       "<td>2.3228745</td>\n",
       "<td>0.8</td>\n",
       "<td>0.7894737</td>\n",
       "<td>0.0256410</td>\n",
       "<td>0.0961538</td>\n",
       "<td>135.3846154</td>\n",
       "<td>132.2874494</td></tr>\n",
       "<tr><td></td>\n",
       "<td>5</td>\n",
       "<td>0.0501089</td>\n",
       "<td>0.8133373</td>\n",
       "<td>2.9423077</td>\n",
       "<td>2.4306020</td>\n",
       "<td>1.0</td>\n",
       "<td>0.8260870</td>\n",
       "<td>0.0256410</td>\n",
       "<td>0.1217949</td>\n",
       "<td>194.2307692</td>\n",
       "<td>143.0602007</td></tr>\n",
       "<tr><td></td>\n",
       "<td>6</td>\n",
       "<td>0.1002179</td>\n",
       "<td>0.7581859</td>\n",
       "<td>2.3026756</td>\n",
       "<td>2.3666388</td>\n",
       "<td>0.7826087</td>\n",
       "<td>0.8043478</td>\n",
       "<td>0.1153846</td>\n",
       "<td>0.2371795</td>\n",
       "<td>130.2675585</td>\n",
       "<td>136.6638796</td></tr>\n",
       "<tr><td></td>\n",
       "<td>7</td>\n",
       "<td>0.1503268</td>\n",
       "<td>0.6984427</td>\n",
       "<td>2.3026756</td>\n",
       "<td>2.3453177</td>\n",
       "<td>0.7826087</td>\n",
       "<td>0.7971014</td>\n",
       "<td>0.1153846</td>\n",
       "<td>0.3525641</td>\n",
       "<td>130.2675585</td>\n",
       "<td>134.5317726</td></tr>\n",
       "<tr><td></td>\n",
       "<td>8</td>\n",
       "<td>0.2004357</td>\n",
       "<td>0.6456386</td>\n",
       "<td>2.0468227</td>\n",
       "<td>2.2706940</td>\n",
       "<td>0.6956522</td>\n",
       "<td>0.7717391</td>\n",
       "<td>0.1025641</td>\n",
       "<td>0.4551282</td>\n",
       "<td>104.6822742</td>\n",
       "<td>127.0693980</td></tr>\n",
       "<tr><td></td>\n",
       "<td>9</td>\n",
       "<td>0.3006536</td>\n",
       "<td>0.4457796</td>\n",
       "<td>1.0234114</td>\n",
       "<td>1.8549331</td>\n",
       "<td>0.3478261</td>\n",
       "<td>0.6304348</td>\n",
       "<td>0.1025641</td>\n",
       "<td>0.5576923</td>\n",
       "<td>2.3411371</td>\n",
       "<td>85.4933110</td></tr>\n",
       "<tr><td></td>\n",
       "<td>10</td>\n",
       "<td>0.4008715</td>\n",
       "<td>0.3251677</td>\n",
       "<td>1.7909699</td>\n",
       "<td>1.8389423</td>\n",
       "<td>0.6086957</td>\n",
       "<td>0.625</td>\n",
       "<td>0.1794872</td>\n",
       "<td>0.7371795</td>\n",
       "<td>79.0969900</td>\n",
       "<td>83.8942308</td></tr>\n",
       "<tr><td></td>\n",
       "<td>11</td>\n",
       "<td>0.5010893</td>\n",
       "<td>0.2363712</td>\n",
       "<td>0.8954849</td>\n",
       "<td>1.6502508</td>\n",
       "<td>0.3043478</td>\n",
       "<td>0.5608696</td>\n",
       "<td>0.0897436</td>\n",
       "<td>0.8269231</td>\n",
       "<td>-10.4515050</td>\n",
       "<td>65.0250836</td></tr>\n",
       "<tr><td></td>\n",
       "<td>12</td>\n",
       "<td>0.5991285</td>\n",
       "<td>0.1800702</td>\n",
       "<td>0.6538462</td>\n",
       "<td>1.4872028</td>\n",
       "<td>0.2222222</td>\n",
       "<td>0.5054545</td>\n",
       "<td>0.0641026</td>\n",
       "<td>0.8910256</td>\n",
       "<td>-34.6153846</td>\n",
       "<td>48.7202797</td></tr>\n",
       "<tr><td></td>\n",
       "<td>13</td>\n",
       "<td>0.6993464</td>\n",
       "<td>0.1429461</td>\n",
       "<td>0.4477425</td>\n",
       "<td>1.3382459</td>\n",
       "<td>0.1521739</td>\n",
       "<td>0.4548287</td>\n",
       "<td>0.0448718</td>\n",
       "<td>0.9358974</td>\n",
       "<td>-55.2257525</td>\n",
       "<td>33.8245866</td></tr>\n",
       "<tr><td></td>\n",
       "<td>14</td>\n",
       "<td>0.7995643</td>\n",
       "<td>0.1124816</td>\n",
       "<td>0.2558528</td>\n",
       "<td>1.2025781</td>\n",
       "<td>0.0869565</td>\n",
       "<td>0.4087193</td>\n",
       "<td>0.0256410</td>\n",
       "<td>0.9615385</td>\n",
       "<td>-74.4147157</td>\n",
       "<td>20.2578076</td></tr>\n",
       "<tr><td></td>\n",
       "<td>15</td>\n",
       "<td>0.8997821</td>\n",
       "<td>0.0903990</td>\n",
       "<td>0.2558528</td>\n",
       "<td>1.0971317</td>\n",
       "<td>0.0869565</td>\n",
       "<td>0.3728814</td>\n",
       "<td>0.0256410</td>\n",
       "<td>0.9871795</td>\n",
       "<td>-74.4147157</td>\n",
       "<td>9.7131682</td></tr>\n",
       "<tr><td></td>\n",
       "<td>16</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0632344</td>\n",
       "<td>0.1279264</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0434783</td>\n",
       "<td>0.3398693</td>\n",
       "<td>0.0128205</td>\n",
       "<td>1.0</td>\n",
       "<td>-87.2073579</td>\n",
       "<td>0.0</td></tr></table></div>"
      ],
      "text/plain": [
       "    group    cumulative_data_fraction    lower_threshold    lift      cumulative_lift    response_rate    cumulative_response_rate    capture_rate    cumulative_capture_rate    gain      cumulative_gain\n",
       "--  -------  --------------------------  -----------------  --------  -----------------  ---------------  --------------------------  --------------  -------------------------  --------  -----------------\n",
       "    1        0.0108932                   0.879095           2.35385   2.35385            0.8              0.8                         0.025641        0.025641                   135.385   135.385\n",
       "    2        0.0217865                   0.856431           2.35385   2.35385            0.8              0.8                         0.025641        0.0512821                  135.385   135.385\n",
       "    3        0.0305011                   0.833303           2.20673   2.31181            0.75             0.785714                    0.0192308       0.0705128                  120.673   131.181\n",
       "    4        0.0413943                   0.829353           2.35385   2.32287            0.8              0.789474                    0.025641        0.0961538                  135.385   132.287\n",
       "    5        0.0501089                   0.813337           2.94231   2.4306             1                0.826087                    0.025641        0.121795                   194.231   143.06\n",
       "    6        0.100218                    0.758186           2.30268   2.36664            0.782609         0.804348                    0.115385        0.237179                   130.268   136.664\n",
       "    7        0.150327                    0.698443           2.30268   2.34532            0.782609         0.797101                    0.115385        0.352564                   130.268   134.532\n",
       "    8        0.200436                    0.645639           2.04682   2.27069            0.695652         0.771739                    0.102564        0.455128                   104.682   127.069\n",
       "    9        0.300654                    0.44578            1.02341   1.85493            0.347826         0.630435                    0.102564        0.557692                   2.34114   85.4933\n",
       "    10       0.400871                    0.325168           1.79097   1.83894            0.608696         0.625                       0.179487        0.737179                   79.097    83.8942\n",
       "    11       0.501089                    0.236371           0.895485  1.65025            0.304348         0.56087                     0.0897436       0.826923                   -10.4515  65.0251\n",
       "    12       0.599129                    0.18007            0.653846  1.4872             0.222222         0.505455                    0.0641026       0.891026                   -34.6154  48.7203\n",
       "    13       0.699346                    0.142946           0.447742  1.33825            0.152174         0.454829                    0.0448718       0.935897                   -55.2258  33.8246\n",
       "    14       0.799564                    0.112482           0.255853  1.20258            0.0869565        0.408719                    0.025641        0.961538                   -74.4147  20.2578\n",
       "    15       0.899782                    0.090399           0.255853  1.09713            0.0869565        0.372881                    0.025641        0.987179                   -74.4147  9.71317\n",
       "    16       1                           0.0632344          0.127926  1                  0.0434783        0.339869                    0.0128205       1                          -87.2074  0"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": []
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "aml.leader"
   ]
  },
  {
   "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": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}