H2O_AutoML_IPYNB/Deep Water Quickstart.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Import Modules"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import h2o\n",
    "import pandas\n",
    "import random\n",
    "from PIL import Image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Populating the interactive namespace from numpy and matplotlib\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.5/dist-packages/IPython/core/magics/pylab.py:160: UserWarning: pylab import has clobbered these variables: ['random']\n",
      "`%matplotlib` prevents importing * from pylab and numpy\n",
      "  \"\\n`%matplotlib` prevents importing * from pylab and numpy\"\n"
     ]
    }
   ],
   "source": [
    "%pylab inline\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.image as mpimg"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "project_path = \"/gtc-2017\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Connect or Start H2O"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Checking whether there is an H2O instance running at http://localhost:54321. connected.\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td>H2O cluster uptime:</td>\n",
       "<td>1 hour 31 mins</td></tr>\n",
       "<tr><td>H2O cluster version:</td>\n",
       "<td>3.11.0.230</td></tr>\n",
       "<tr><td>H2O cluster version age:</td>\n",
       "<td>15 hours and 17 minutes </td></tr>\n",
       "<tr><td>H2O cluster name:</td>\n",
       "<td>root</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>25.75 Gb</td></tr>\n",
       "<tr><td>H2O cluster total cores:</td>\n",
       "<td>20</td></tr>\n",
       "<tr><td>H2O cluster allowed cores:</td>\n",
       "<td>20</td></tr>\n",
       "<tr><td>H2O cluster status:</td>\n",
       "<td>locked, healthy</td></tr>\n",
       "<tr><td>H2O connection url:</td>\n",
       "<td>http://localhost:54321</td></tr>\n",
       "<tr><td>H2O connection proxy:</td>\n",
       "<td>None</td></tr>\n",
       "<tr><td>H2O internal security:</td>\n",
       "<td>False</td></tr>\n",
       "<tr><td>Python version:</td>\n",
       "<td>3.5.2 final</td></tr></table></div>"
      ],
      "text/plain": [
       "--------------------------  -----------------------\n",
       "H2O cluster uptime:         1 hour 31 mins\n",
       "H2O cluster version:        3.11.0.230\n",
       "H2O cluster version age:    15 hours and 17 minutes\n",
       "H2O cluster name:           root\n",
       "H2O cluster total nodes:    1\n",
       "H2O cluster free memory:    25.75 Gb\n",
       "H2O cluster total cores:    20\n",
       "H2O cluster allowed cores:  20\n",
       "H2O cluster status:         locked, healthy\n",
       "H2O connection url:         http://localhost:54321\n",
       "H2O connection proxy:\n",
       "H2O internal security:      False\n",
       "Python version:             3.5.2 final\n",
       "--------------------------  -----------------------"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "h2o.init()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Import Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Import data into H2O."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n",
      "Parse progress: |█████████████████████████████████████████████████████████| 100%\n"
     ]
    }
   ],
   "source": [
    "mnist_training = h2o.import_file(project_path+\"/data/mnist-training.csv\")\n",
    "mnist_testing = h2o.import_file(project_path+\"/data/mnist-testing.csv\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Change the \"label\" column to a factor (i.e. categorical feature)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "mnist_training[\"label\"] = mnist_training[\"label\"].asfactor()\n",
    "mnist_testing[\"label\"] = mnist_testing[\"label\"].asfactor()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Explore Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table>\n",
       "<thead>\n",
       "<tr><th>uri                                          </th><th style=\"text-align: right;\">  label</th></tr>\n",
       "</thead>\n",
       "<tbody>\n",
       "<tr><td>/gtc-2017/data/mnist_png/training/6/6453.png </td><td style=\"text-align: right;\">      6</td></tr>\n",
       "<tr><td>/gtc-2017/data/mnist_png/training/6/13970.png</td><td style=\"text-align: right;\">      6</td></tr>\n",
       "<tr><td>/gtc-2017/data/mnist_png/training/6/21967.png</td><td style=\"text-align: right;\">      6</td></tr>\n",
       "<tr><td>/gtc-2017/data/mnist_png/training/6/11277.png</td><td style=\"text-align: right;\">      6</td></tr>\n",
       "<tr><td>/gtc-2017/data/mnist_png/training/6/27350.png</td><td style=\"text-align: right;\">      6</td></tr>\n",
       "<tr><td>/gtc-2017/data/mnist_png/training/6/32119.png</td><td style=\"text-align: right;\">      6</td></tr>\n",
       "<tr><td>/gtc-2017/data/mnist_png/training/6/44201.png</td><td style=\"text-align: right;\">      6</td></tr>\n",
       "<tr><td>/gtc-2017/data/mnist_png/training/6/58770.png</td><td style=\"text-align: right;\">      6</td></tr>\n",
       "<tr><td>/gtc-2017/data/mnist_png/training/6/39440.png</td><td style=\"text-align: right;\">      6</td></tr>\n",
       "<tr><td>/gtc-2017/data/mnist_png/training/6/41988.png</td><td style=\"text-align: right;\">      6</td></tr>\n",
       "</tbody>\n",
       "</table>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": []
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mnist_training.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's view a sample fo the images."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "pd_mnist_training = mnist_training.as_data_frame()\n",
    "list_image_uris = pd_mnist_training[\"uri\"].tolist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f058d59c320>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSIrwA0kRfiApwg8kRfiBpAg/kNT/A3c7BqkZfRWPAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f058d62a160>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nCD8Q1L8AAYk87Rp9rwsAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f058d5a0d30>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\naFf8AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f05b071a5c0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f058d5a0b70>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f058d8fcb70>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nEX4gKcIPJEX4gaQIP5AU4QeS+j+FCuimvuJumAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f058d6de518>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n8ANBEX4gqP8HlsKJWwHV+2sAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f058da01320>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nwg8ERfiBoAg/EBThB4L6H24eGo770a29AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f058d7a5a58>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f058d8a2ef0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "num_sample_images = 10\n",
    "\n",
    "for x in range(num_sample_images):\n",
    "    uri = random.choice(list_image_uris)\n",
    "    img=mpimg.imread(uri)\n",
    "    imgplot = plt.imshow(img, cmap=\"Greys\")\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Test GPU Driver"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tue May  9 04:51:16 2017       \r\n",
      "+-----------------------------------------------------------------------------+\r\n",
      "| NVIDIA-SMI 378.13                 Driver Version: 378.13                    |\r\n",
      "|-------------------------------+----------------------+----------------------+\r\n",
      "| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |\r\n",
      "| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |\r\n",
      "|===============================+======================+======================|\r\n",
      "|   0  GeForce GTX 1080    Off  | 0000:02:00.0      On |                  N/A |\r\n",
      "| 30%   39C    P8    11W / 180W |   4799MiB /  8112MiB |      2%      Default |\r\n",
      "+-------------------------------+----------------------+----------------------+\r\n",
      "                                                                               \r\n",
      "+-----------------------------------------------------------------------------+\r\n",
      "| Processes:                                                       GPU Memory |\r\n",
      "|  GPU       PID  Type  Process name                               Usage      |\r\n",
      "|=============================================================================|\r\n",
      "+-----------------------------------------------------------------------------+\r\n"
     ]
    }
   ],
   "source": [
    "!nvidia-smi"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Build Deep Water MXNet Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The MXNet backend is the default, so if no backend is explicitly set, MXNet will be used."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "deepwater Model Build progress: |█████████████████████████████████████████| 100%\n"
     ]
    }
   ],
   "source": [
    "from h2o.estimators.deepwater import H2ODeepWaterEstimator\n",
    "model_mnist_lenet_mx = H2ODeepWaterEstimator(epochs=80, network=\"lenet\")\n",
    "model_mnist_lenet_mx.train(x=[\"uri\"], y=\"label\",\n",
    "                           training_frame=mnist_training,\n",
    "                           validation_frame=mnist_testing,\n",
    "                           model_id=\"model_mnist_lenet_mx\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model Details\n",
      "=============\n",
      "H2ODeepWaterEstimator :  Deep Water\n",
      "Model Key:  model_mnist_lenet_mx\n",
      "\n",
      "\n",
      "ModelMetricsMultinomial: deepwater\n",
      "** Reported on train data. **\n",
      "\n",
      "MSE: 2.9238779147804818e-05\n",
      "RMSE: 0.005407289445535981\n",
      "LogLoss: 0.00022719911928558199\n",
      "Mean Per-Class Error: 0.0\n",
      "Confusion Matrix: vertical: actual; across: predicted\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b>0</b></td>\n",
       "<td><b>1</b></td>\n",
       "<td><b>2</b></td>\n",
       "<td><b>3</b></td>\n",
       "<td><b>4</b></td>\n",
       "<td><b>5</b></td>\n",
       "<td><b>6</b></td>\n",
       "<td><b>7</b></td>\n",
       "<td><b>8</b></td>\n",
       "<td><b>9</b></td>\n",
       "<td><b>Error</b></td>\n",
       "<td><b>Rate</b></td></tr>\n",
       "<tr><td>938.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0 / 938</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>1103.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0 / 1,103</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1026.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0 / 1,026</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>994.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0 / 994</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>976.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0 / 976</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>898.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0 / 898</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1002.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0 / 1,002</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1055.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0 / 1,055</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>926.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0 / 926</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1004.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0 / 1,004</td></tr>\n",
       "<tr><td>938.0</td>\n",
       "<td>1103.0</td>\n",
       "<td>1026.0</td>\n",
       "<td>994.0</td>\n",
       "<td>976.0</td>\n",
       "<td>898.0</td>\n",
       "<td>1002.0</td>\n",
       "<td>1055.0</td>\n",
       "<td>926.0</td>\n",
       "<td>1004.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0 / 9,922</td></tr></table></div>"
      ],
      "text/plain": [
       "0    1     2     3    4    5    6     7     8    9     Error    Rate\n",
       "---  ----  ----  ---  ---  ---  ----  ----  ---  ----  -------  ---------\n",
       "938  0     0     0    0    0    0     0     0    0     0        0 / 938\n",
       "0    1103  0     0    0    0    0     0     0    0     0        0 / 1,103\n",
       "0    0     1026  0    0    0    0     0     0    0     0        0 / 1,026\n",
       "0    0     0     994  0    0    0     0     0    0     0        0 / 994\n",
       "0    0     0     0    976  0    0     0     0    0     0        0 / 976\n",
       "0    0     0     0    0    898  0     0     0    0     0        0 / 898\n",
       "0    0     0     0    0    0    1002  0     0    0     0        0 / 1,002\n",
       "0    0     0     0    0    0    0     1055  0    0     0        0 / 1,055\n",
       "0    0     0     0    0    0    0     0     926  0     0        0 / 926\n",
       "0    0     0     0    0    0    0     0     0    1004  0        0 / 1,004\n",
       "938  1103  1026  994  976  898  1002  1055  926  1004  0        0 / 9,922"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Top-10 Hit Ratios: \n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b>k</b></td>\n",
       "<td><b>hit_ratio</b></td></tr>\n",
       "<tr><td>1</td>\n",
       "<td>1.0</td></tr>\n",
       "<tr><td>2</td>\n",
       "<td>1.0</td></tr>\n",
       "<tr><td>3</td>\n",
       "<td>1.0</td></tr>\n",
       "<tr><td>4</td>\n",
       "<td>1.0</td></tr>\n",
       "<tr><td>5</td>\n",
       "<td>1.0</td></tr>\n",
       "<tr><td>6</td>\n",
       "<td>1.0</td></tr>\n",
       "<tr><td>7</td>\n",
       "<td>1.0</td></tr>\n",
       "<tr><td>8</td>\n",
       "<td>1.0</td></tr>\n",
       "<tr><td>9</td>\n",
       "<td>1.0</td></tr>\n",
       "<tr><td>10</td>\n",
       "<td>1.0</td></tr></table></div>"
      ],
      "text/plain": [
       "k    hit_ratio\n",
       "---  -----------\n",
       "1    1\n",
       "2    1\n",
       "3    1\n",
       "4    1\n",
       "5    1\n",
       "6    1\n",
       "7    1\n",
       "8    1\n",
       "9    1\n",
       "10   1"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "ModelMetricsMultinomial: deepwater\n",
      "** Reported on validation data. **\n",
      "\n",
      "MSE: 0.009023165097423982\n",
      "RMSE: 0.09499034212710249\n",
      "LogLoss: 0.05017128162863757\n",
      "Mean Per-Class Error: 0.01078546363714374\n",
      "Confusion Matrix: vertical: actual; across: predicted\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b>0</b></td>\n",
       "<td><b>1</b></td>\n",
       "<td><b>2</b></td>\n",
       "<td><b>3</b></td>\n",
       "<td><b>4</b></td>\n",
       "<td><b>5</b></td>\n",
       "<td><b>6</b></td>\n",
       "<td><b>7</b></td>\n",
       "<td><b>8</b></td>\n",
       "<td><b>9</b></td>\n",
       "<td><b>Error</b></td>\n",
       "<td><b>Rate</b></td></tr>\n",
       "<tr><td>973.0</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>3.0</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0071429</td>\n",
       "<td>7 / 980</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>1132.0</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0026432</td>\n",
       "<td>3 / 1,135</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>2.0</td>\n",
       "<td>1022.0</td>\n",
       "<td>2.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>4.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0096899</td>\n",
       "<td>10 / 1,032</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1006.0</td>\n",
       "<td>0.0</td>\n",
       "<td>2.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0039604</td>\n",
       "<td>4 / 1,010</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>969.0</td>\n",
       "<td>0.0</td>\n",
       "<td>2.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>10.0</td>\n",
       "<td>0.0132383</td>\n",
       "<td>13 / 982</td></tr>\n",
       "<tr><td>2.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>6.0</td>\n",
       "<td>0.0</td>\n",
       "<td>879.0</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>2.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0145740</td>\n",
       "<td>13 / 892</td></tr>\n",
       "<tr><td>3.0</td>\n",
       "<td>4.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>3.0</td>\n",
       "<td>5.0</td>\n",
       "<td>941.0</td>\n",
       "<td>0.0</td>\n",
       "<td>2.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0177453</td>\n",
       "<td>17 / 958</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>2.0</td>\n",
       "<td>3.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1018.0</td>\n",
       "<td>1.0</td>\n",
       "<td>2.0</td>\n",
       "<td>0.0097276</td>\n",
       "<td>10 / 1,028</td></tr>\n",
       "<tr><td>3.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>2.0</td>\n",
       "<td>0.0</td>\n",
       "<td>2.0</td>\n",
       "<td>963.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0112936</td>\n",
       "<td>11 / 974</td></tr>\n",
       "<tr><td>1.0</td>\n",
       "<td>3.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>7.0</td>\n",
       "<td>3.0</td>\n",
       "<td>0.0</td>\n",
       "<td>3.0</td>\n",
       "<td>0.0</td>\n",
       "<td>991.0</td>\n",
       "<td>0.0178394</td>\n",
       "<td>18 / 1,009</td></tr>\n",
       "<tr><td>982.0</td>\n",
       "<td>1144.0</td>\n",
       "<td>1029.0</td>\n",
       "<td>1016.0</td>\n",
       "<td>981.0</td>\n",
       "<td>892.0</td>\n",
       "<td>948.0</td>\n",
       "<td>1031.0</td>\n",
       "<td>971.0</td>\n",
       "<td>1006.0</td>\n",
       "<td>0.0106</td>\n",
       "<td>106 / 10,000</td></tr></table></div>"
      ],
      "text/plain": [
       "0    1     2     3     4    5    6    7     8    9     Error       Rate\n",
       "---  ----  ----  ----  ---  ---  ---  ----  ---  ----  ----------  ------------\n",
       "973  1     1     0     0    0    3    1     1    0     0.00714286  7 / 980\n",
       "0    1132  1     1     0    0    0    1     0    0     0.00264317  3 / 1,135\n",
       "0    2     1022  2     0    0    1    4     1    0     0.00968992  10 / 1,032\n",
       "0    0     0     1006  0    2    0    0     1    1     0.0039604   4 / 1,010\n",
       "0    0     0     0     969  0    2    1     0    10    0.0132383   13 / 982\n",
       "2    0     0     6     0    879  1    1     2    1     0.014574    13 / 892\n",
       "3    4     0     0     3    5    941  0     2    0     0.0177453   17 / 958\n",
       "0    2     3     0     1    1    0    1018  1    2     0.00972763  10 / 1,028\n",
       "3    0     1     1     1    2    0    2     963  1     0.0112936   11 / 974\n",
       "1    3     1     0     7    3    0    3     0    991   0.0178394   18 / 1,009\n",
       "982  1144  1029  1016  981  892  948  1031  971  1006  0.0106      106 / 10,000"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Top-10 Hit Ratios: \n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b>k</b></td>\n",
       "<td><b>hit_ratio</b></td></tr>\n",
       "<tr><td>1</td>\n",
       "<td>0.9894</td></tr>\n",
       "<tr><td>2</td>\n",
       "<td>0.9988000</td></tr>\n",
       "<tr><td>3</td>\n",
       "<td>0.9997</td></tr>\n",
       "<tr><td>4</td>\n",
       "<td>0.9998</td></tr>\n",
       "<tr><td>5</td>\n",
       "<td>0.9999000</td></tr>\n",
       "<tr><td>6</td>\n",
       "<td>1.0</td></tr>\n",
       "<tr><td>7</td>\n",
       "<td>1.0</td></tr>\n",
       "<tr><td>8</td>\n",
       "<td>1.0</td></tr>\n",
       "<tr><td>9</td>\n",
       "<td>1.0</td></tr>\n",
       "<tr><td>10</td>\n",
       "<td>1.0</td></tr></table></div>"
      ],
      "text/plain": [
       "k    hit_ratio\n",
       "---  -----------\n",
       "1    0.9894\n",
       "2    0.9988\n",
       "3    0.9997\n",
       "4    0.9998\n",
       "5    0.9999\n",
       "6    1\n",
       "7    1\n",
       "8    1\n",
       "9    1\n",
       "10   1"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "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>training_speed</b></td>\n",
       "<td><b>epochs</b></td>\n",
       "<td><b>iterations</b></td>\n",
       "<td><b>samples</b></td>\n",
       "<td><b>training_rmse</b></td>\n",
       "<td><b>training_logloss</b></td>\n",
       "<td><b>training_classification_error</b></td>\n",
       "<td><b>validation_rmse</b></td>\n",
       "<td><b>validation_logloss</b></td>\n",
       "<td><b>validation_classification_error</b></td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:51:16</td>\n",
       "<td> 0.000 sec</td>\n",
       "<td>None</td>\n",
       "<td>0.0</td>\n",
       "<td>0</td>\n",
       "<td>0.0</td>\n",
       "<td>nan</td>\n",
       "<td>nan</td>\n",
       "<td>nan</td>\n",
       "<td>nan</td>\n",
       "<td>nan</td>\n",
       "<td>nan</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:51:16</td>\n",
       "<td> 0.951 sec</td>\n",
       "<td>11130 obs/sec</td>\n",
       "<td>0.0170667</td>\n",
       "<td>1</td>\n",
       "<td>1024.0</td>\n",
       "<td>0.5083504</td>\n",
       "<td>2.1276716</td>\n",
       "<td>0.2816972</td>\n",
       "<td>0.5052221</td>\n",
       "<td>2.0969191</td>\n",
       "<td>0.2767</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:51:25</td>\n",
       "<td> 9.022 sec</td>\n",
       "<td>18467 obs/sec</td>\n",
       "<td>2.3552</td>\n",
       "<td>138</td>\n",
       "<td>141312.0</td>\n",
       "<td>0.1286586</td>\n",
       "<td>0.0692300</td>\n",
       "<td>0.0199557</td>\n",
       "<td>0.1514920</td>\n",
       "<td>0.0989359</td>\n",
       "<td>0.027</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:51:30</td>\n",
       "<td>14.175 sec</td>\n",
       "<td>18358 obs/sec</td>\n",
       "<td>3.7717333</td>\n",
       "<td>221</td>\n",
       "<td>226304.0</td>\n",
       "<td>0.0833549</td>\n",
       "<td>0.0300278</td>\n",
       "<td>0.0081637</td>\n",
       "<td>0.1115630</td>\n",
       "<td>0.0575506</td>\n",
       "<td>0.0148</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:51:35</td>\n",
       "<td>19.188 sec</td>\n",
       "<td>18054 obs/sec</td>\n",
       "<td>5.0858667</td>\n",
       "<td>298</td>\n",
       "<td>305152.0</td>\n",
       "<td>0.0666445</td>\n",
       "<td>0.0175441</td>\n",
       "<td>0.0047369</td>\n",
       "<td>0.1106728</td>\n",
       "<td>0.0569475</td>\n",
       "<td>0.0143</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:51:40</td>\n",
       "<td>24.297 sec</td>\n",
       "<td>17637 obs/sec</td>\n",
       "<td>6.3317333</td>\n",
       "<td>371</td>\n",
       "<td>379904.0</td>\n",
       "<td>0.0562053</td>\n",
       "<td>0.0120473</td>\n",
       "<td>0.0039307</td>\n",
       "<td>0.1071403</td>\n",
       "<td>0.0646352</td>\n",
       "<td>0.0131</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:51:45</td>\n",
       "<td>29.392 sec</td>\n",
       "<td>17427 obs/sec</td>\n",
       "<td>7.5946667</td>\n",
       "<td>445</td>\n",
       "<td>455680.0</td>\n",
       "<td>0.0254586</td>\n",
       "<td>0.0023757</td>\n",
       "<td>0.0010079</td>\n",
       "<td>0.1008606</td>\n",
       "<td>0.0519758</td>\n",
       "<td>0.0123</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:51:50</td>\n",
       "<td>34.493 sec</td>\n",
       "<td>17258 obs/sec</td>\n",
       "<td>8.8405333</td>\n",
       "<td>518</td>\n",
       "<td>530432.0</td>\n",
       "<td>0.0219109</td>\n",
       "<td>0.0019636</td>\n",
       "<td>0.0005039</td>\n",
       "<td>0.1000205</td>\n",
       "<td>0.0536413</td>\n",
       "<td>0.0113</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:51:55</td>\n",
       "<td>39.695 sec</td>\n",
       "<td>17138 obs/sec</td>\n",
       "<td>10.1034667</td>\n",
       "<td>592</td>\n",
       "<td>606208.0</td>\n",
       "<td>0.0117912</td>\n",
       "<td>0.0006531</td>\n",
       "<td>0.0001008</td>\n",
       "<td>0.0966056</td>\n",
       "<td>0.0507177</td>\n",
       "<td>0.0111</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:52:01</td>\n",
       "<td>45.341 sec</td>\n",
       "<td>17088 obs/sec</td>\n",
       "<td>11.5370667</td>\n",
       "<td>676</td>\n",
       "<td>692224.0</td>\n",
       "<td>0.0054073</td>\n",
       "<td>0.0002272</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0949903</td>\n",
       "<td>0.0501713</td>\n",
       "<td>0.0106</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:52:06</td>\n",
       "<td>50.481 sec</td>\n",
       "<td>16989 obs/sec</td>\n",
       "<td>12.7829333</td>\n",
       "<td>749</td>\n",
       "<td>766976.0</td>\n",
       "<td>0.0029129</td>\n",
       "<td>0.0001506</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0960029</td>\n",
       "<td>0.0523594</td>\n",
       "<td>0.0107</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:52:11</td>\n",
       "<td>55.577 sec</td>\n",
       "<td>16923 obs/sec</td>\n",
       "<td>14.0288</td>\n",
       "<td>822</td>\n",
       "<td>841728.0</td>\n",
       "<td>0.0012404</td>\n",
       "<td>0.0000801</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0947228</td>\n",
       "<td>0.0507274</td>\n",
       "<td>0.0108</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:52:16</td>\n",
       "<td> 1 min  0.607 sec</td>\n",
       "<td>16862 obs/sec</td>\n",
       "<td>15.2576</td>\n",
       "<td>894</td>\n",
       "<td>915456.0</td>\n",
       "<td>0.0006823</td>\n",
       "<td>0.0000591</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0945490</td>\n",
       "<td>0.0511729</td>\n",
       "<td>0.0106</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:52:21</td>\n",
       "<td> 1 min  5.624 sec</td>\n",
       "<td>16827 obs/sec</td>\n",
       "<td>16.5034667</td>\n",
       "<td>967</td>\n",
       "<td>990208.0</td>\n",
       "<td>0.0006221</td>\n",
       "<td>0.0000517</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0943962</td>\n",
       "<td>0.0512009</td>\n",
       "<td>0.0104</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:52:26</td>\n",
       "<td> 1 min 10.661 sec</td>\n",
       "<td>16795 obs/sec</td>\n",
       "<td>17.7493333</td>\n",
       "<td>1040</td>\n",
       "<td>1064960.0</td>\n",
       "<td>0.0005839</td>\n",
       "<td>0.0000473</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0944581</td>\n",
       "<td>0.0516068</td>\n",
       "<td>0.0103</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:52:31</td>\n",
       "<td> 1 min 15.712 sec</td>\n",
       "<td>16723 obs/sec</td>\n",
       "<td>18.944</td>\n",
       "<td>1110</td>\n",
       "<td>1136640.0</td>\n",
       "<td>0.0004731</td>\n",
       "<td>0.0000422</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0947969</td>\n",
       "<td>0.0525148</td>\n",
       "<td>0.0107</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:52:36</td>\n",
       "<td> 1 min 20.732 sec</td>\n",
       "<td>16681 obs/sec</td>\n",
       "<td>20.1557333</td>\n",
       "<td>1181</td>\n",
       "<td>1209344.0</td>\n",
       "<td>0.0004513</td>\n",
       "<td>0.0000402</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0943542</td>\n",
       "<td>0.0523436</td>\n",
       "<td>0.0106</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:52:41</td>\n",
       "<td> 1 min 25.816 sec</td>\n",
       "<td>16654 obs/sec</td>\n",
       "<td>21.4016</td>\n",
       "<td>1254</td>\n",
       "<td>1284096.0</td>\n",
       "<td>0.0004102</td>\n",
       "<td>0.0000369</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0941857</td>\n",
       "<td>0.0527283</td>\n",
       "<td>0.0102</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:52:42</td>\n",
       "<td> 1 min 26.296 sec</td>\n",
       "<td>16650 obs/sec</td>\n",
       "<td>21.4016</td>\n",
       "<td>1254</td>\n",
       "<td>1284096.0</td>\n",
       "<td>0.0054073</td>\n",
       "<td>0.0002272</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0949903</td>\n",
       "<td>0.0501713</td>\n",
       "<td>0.0106</td></tr></table></div>"
      ],
      "text/plain": [
       "    timestamp            duration          training_speed    epochs     iterations    samples      training_rmse    training_logloss    training_classification_error    validation_rmse    validation_logloss    validation_classification_error\n",
       "--  -------------------  ----------------  ----------------  ---------  ------------  -----------  ---------------  ------------------  -------------------------------  -----------------  --------------------  ---------------------------------\n",
       "    2017-05-09 04:51:16  0.000 sec                           0          0             0            nan              nan                 nan                              nan                nan                   nan\n",
       "    2017-05-09 04:51:16  0.951 sec         11130 obs/sec     0.0170667  1             1024         0.50835          2.12767             0.281697                         0.505222           2.09692               0.2767\n",
       "    2017-05-09 04:51:25  9.022 sec         18467 obs/sec     2.3552     138           141312       0.128659         0.06923             0.0199557                        0.151492           0.0989359             0.027\n",
       "    2017-05-09 04:51:30  14.175 sec        18358 obs/sec     3.77173    221           226304       0.0833549        0.0300278           0.00816368                       0.111563           0.0575506             0.0148\n",
       "    2017-05-09 04:51:35  19.188 sec        18054 obs/sec     5.08587    298           305152       0.0666445        0.0175441           0.00473695                       0.110673           0.0569475             0.0143\n",
       "    2017-05-09 04:51:40  24.297 sec        17637 obs/sec     6.33173    371           379904       0.0562053        0.0120473           0.00393066                       0.10714            0.0646352             0.0131\n",
       "    2017-05-09 04:51:45  29.392 sec        17427 obs/sec     7.59467    445           455680       0.0254586        0.00237575          0.00100786                       0.100861           0.0519758             0.0123\n",
       "    2017-05-09 04:51:50  34.493 sec        17258 obs/sec     8.84053    518           530432       0.0219109        0.00196359          0.000503931                      0.10002            0.0536413             0.0113\n",
       "    2017-05-09 04:51:55  39.695 sec        17138 obs/sec     10.1035    592           606208       0.0117912        0.000653079         0.000100786                      0.0966056          0.0507177             0.0111\n",
       "    2017-05-09 04:52:01  45.341 sec        17088 obs/sec     11.5371    676           692224       0.00540729       0.000227199         0                                0.0949903          0.0501713             0.0106\n",
       "    2017-05-09 04:52:06  50.481 sec        16989 obs/sec     12.7829    749           766976       0.00291287       0.000150645         0                                0.0960029          0.0523594             0.0107\n",
       "    2017-05-09 04:52:11  55.577 sec        16923 obs/sec     14.0288    822           841728       0.00124043       8.00647e-05         0                                0.0947228          0.0507274             0.0108\n",
       "    2017-05-09 04:52:16  1 min  0.607 sec  16862 obs/sec     15.2576    894           915456       0.000682257      5.91092e-05         0                                0.094549           0.0511729             0.0106\n",
       "    2017-05-09 04:52:21  1 min  5.624 sec  16827 obs/sec     16.5035    967           990208       0.00062214       5.17277e-05         0                                0.0943962          0.0512009             0.0104\n",
       "    2017-05-09 04:52:26  1 min 10.661 sec  16795 obs/sec     17.7493    1040          1.06496e+06  0.000583933      4.73307e-05         0                                0.0944581          0.0516068             0.0103\n",
       "    2017-05-09 04:52:31  1 min 15.712 sec  16723 obs/sec     18.944     1110          1.13664e+06  0.000473093      4.21779e-05         0                                0.0947969          0.0525148             0.0107\n",
       "    2017-05-09 04:52:36  1 min 20.732 sec  16681 obs/sec     20.1557    1181          1.20934e+06  0.00045129       4.01928e-05         0                                0.0943542          0.0523436             0.0106\n",
       "    2017-05-09 04:52:41  1 min 25.816 sec  16654 obs/sec     21.4016    1254          1.2841e+06   0.000410246      3.69458e-05         0                                0.0941857          0.0527283             0.0102\n",
       "    2017-05-09 04:52:42  1 min 26.296 sec  16650 obs/sec     21.4016    1254          1.2841e+06   0.00540729       0.000227199         0                                0.0949903          0.0501713             0.0106"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model_mnist_lenet_mx.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Build Deep Water TensorFlow Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "deepwater Model Build progress: |█████████████████████████████████████████| 100%\n"
     ]
    }
   ],
   "source": [
    "model_mnist_lenet_tf = H2ODeepWaterEstimator(epochs=80, network=\"lenet\", backend=\"tensorflow\")\n",
    "model_mnist_lenet_tf.train(x=[\"uri\"], y=\"label\",\n",
    "                           training_frame=mnist_training,\n",
    "                           validation_frame=mnist_testing,\n",
    "                           model_id=\"model_mnist_lenet_tf\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model Details\n",
      "=============\n",
      "H2ODeepWaterEstimator :  Deep Water\n",
      "Model Key:  model_mnist_lenet_tf\n",
      "\n",
      "\n",
      "ModelMetricsMultinomial: deepwater\n",
      "** Reported on train data. **\n",
      "\n",
      "MSE: 0.008807465506858264\n",
      "RMSE: 0.09384809804603535\n",
      "LogLoss: 0.03852629216180632\n",
      "Mean Per-Class Error: 0.010554580973143171\n",
      "Confusion Matrix: vertical: actual; across: predicted\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b>0</b></td>\n",
       "<td><b>1</b></td>\n",
       "<td><b>2</b></td>\n",
       "<td><b>3</b></td>\n",
       "<td><b>4</b></td>\n",
       "<td><b>5</b></td>\n",
       "<td><b>6</b></td>\n",
       "<td><b>7</b></td>\n",
       "<td><b>8</b></td>\n",
       "<td><b>9</b></td>\n",
       "<td><b>Error</b></td>\n",
       "<td><b>Rate</b></td></tr>\n",
       "<tr><td>991.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>2.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0040201</td>\n",
       "<td>4 / 995</td></tr>\n",
       "<tr><td>3.0</td>\n",
       "<td>1108.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0053860</td>\n",
       "<td>6 / 1,114</td></tr>\n",
       "<tr><td>1.0</td>\n",
       "<td>3.0</td>\n",
       "<td>991.0</td>\n",
       "<td>5.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0129482</td>\n",
       "<td>13 / 1,004</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>2.0</td>\n",
       "<td>1008.0</td>\n",
       "<td>0.0</td>\n",
       "<td>3.0</td>\n",
       "<td>0.0</td>\n",
       "<td>2.0</td>\n",
       "<td>4.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0117647</td>\n",
       "<td>12 / 1,020</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>966.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>10.0</td>\n",
       "<td>0.0112590</td>\n",
       "<td>11 / 977</td></tr>\n",
       "<tr><td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>894.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0044543</td>\n",
       "<td>4 / 898</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>4.0</td>\n",
       "<td>1023.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0048638</td>\n",
       "<td>5 / 1,028</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>3.0</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>2.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1015.0</td>\n",
       "<td>0.0</td>\n",
       "<td>18.0</td>\n",
       "<td>0.0240385</td>\n",
       "<td>25 / 1,040</td></tr>\n",
       "<tr><td>2.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>7.0</td>\n",
       "<td>5.0</td>\n",
       "<td>0.0</td>\n",
       "<td>947.0</td>\n",
       "<td>3.0</td>\n",
       "<td>0.0206825</td>\n",
       "<td>20 / 967</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>4.0</td>\n",
       "<td>973.0</td>\n",
       "<td>0.0061287</td>\n",
       "<td>6 / 979</td></tr>\n",
       "<tr><td>998.0</td>\n",
       "<td>1115.0</td>\n",
       "<td>994.0</td>\n",
       "<td>1017.0</td>\n",
       "<td>970.0</td>\n",
       "<td>911.0</td>\n",
       "<td>1031.0</td>\n",
       "<td>1020.0</td>\n",
       "<td>960.0</td>\n",
       "<td>1006.0</td>\n",
       "<td>0.0105767</td>\n",
       "<td>106 / 10,022</td></tr></table></div>"
      ],
      "text/plain": [
       "0    1     2    3     4    5    6     7     8    9     Error       Rate\n",
       "---  ----  ---  ----  ---  ---  ----  ----  ---  ----  ----------  ------------\n",
       "991  0     0    0     0    2    0     0     1    1     0.0040201   4 / 995\n",
       "3    1108  0    0     0    1    0     1     1    0     0.005386    6 / 1,114\n",
       "1    3     991  5     1    0    1     1     1    0     0.0129482   13 / 1,004\n",
       "0    0     2    1008  0    3    0     2     4    1     0.0117647   12 / 1,020\n",
       "0    0     0    0     966  0    0     1     0    10    0.011259    11 / 977\n",
       "1    0     0    1     0    894  1     0     1    0     0.00445434  4 / 898\n",
       "0    0     0    0     0    4    1023  0     1    0     0.00486381  5 / 1,028\n",
       "0    3     1    1     2    0    0     1015  0    18    0.0240385   25 / 1,040\n",
       "2    1     0    1     1    7    5     0     947  3     0.0206825   20 / 967\n",
       "0    0     0    1     0    0    1     0     4    973   0.0061287   6 / 979\n",
       "998  1115  994  1017  970  911  1031  1020  960  1006  0.0105767   106 / 10,022"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Top-10 Hit Ratios: \n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b>k</b></td>\n",
       "<td><b>hit_ratio</b></td></tr>\n",
       "<tr><td>1</td>\n",
       "<td>0.9894233</td></tr>\n",
       "<tr><td>2</td>\n",
       "<td>0.9982039</td></tr>\n",
       "<tr><td>3</td>\n",
       "<td>0.9997007</td></tr>\n",
       "<tr><td>4</td>\n",
       "<td>0.9999002</td></tr>\n",
       "<tr><td>5</td>\n",
       "<td>1.0</td></tr>\n",
       "<tr><td>6</td>\n",
       "<td>1.0</td></tr>\n",
       "<tr><td>7</td>\n",
       "<td>1.0</td></tr>\n",
       "<tr><td>8</td>\n",
       "<td>1.0</td></tr>\n",
       "<tr><td>9</td>\n",
       "<td>1.0</td></tr>\n",
       "<tr><td>10</td>\n",
       "<td>1.0</td></tr></table></div>"
      ],
      "text/plain": [
       "k    hit_ratio\n",
       "---  -----------\n",
       "1    0.989423\n",
       "2    0.998204\n",
       "3    0.999701\n",
       "4    0.9999\n",
       "5    1\n",
       "6    1\n",
       "7    1\n",
       "8    1\n",
       "9    1\n",
       "10   1"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "ModelMetricsMultinomial: deepwater\n",
      "** Reported on validation data. **\n",
      "\n",
      "MSE: 0.017004871022782454\n",
      "RMSE: 0.13040272628585053\n",
      "LogLoss: 0.08582499693605794\n",
      "Mean Per-Class Error: 0.018464224835674437\n",
      "Confusion Matrix: vertical: actual; across: predicted\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b>0</b></td>\n",
       "<td><b>1</b></td>\n",
       "<td><b>2</b></td>\n",
       "<td><b>3</b></td>\n",
       "<td><b>4</b></td>\n",
       "<td><b>5</b></td>\n",
       "<td><b>6</b></td>\n",
       "<td><b>7</b></td>\n",
       "<td><b>8</b></td>\n",
       "<td><b>9</b></td>\n",
       "<td><b>Error</b></td>\n",
       "<td><b>Rate</b></td></tr>\n",
       "<tr><td>971.0</td>\n",
       "<td>0.0</td>\n",
       "<td>3.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>2.0</td>\n",
       "<td>2.0</td>\n",
       "<td>0.0</td>\n",
       "<td>2.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0091837</td>\n",
       "<td>9 / 980</td></tr>\n",
       "<tr><td>4.0</td>\n",
       "<td>1126.0</td>\n",
       "<td>2.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>3.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0079295</td>\n",
       "<td>9 / 1,135</td></tr>\n",
       "<tr><td>1.0</td>\n",
       "<td>2.0</td>\n",
       "<td>1020.0</td>\n",
       "<td>3.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>2.0</td>\n",
       "<td>2.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0116279</td>\n",
       "<td>12 / 1,032</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>2.0</td>\n",
       "<td>993.0</td>\n",
       "<td>0.0</td>\n",
       "<td>6.0</td>\n",
       "<td>0.0</td>\n",
       "<td>3.0</td>\n",
       "<td>5.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0168317</td>\n",
       "<td>17 / 1,010</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>2.0</td>\n",
       "<td>0.0</td>\n",
       "<td>966.0</td>\n",
       "<td>0.0</td>\n",
       "<td>3.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>9.0</td>\n",
       "<td>0.0162933</td>\n",
       "<td>16 / 982</td></tr>\n",
       "<tr><td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>6.0</td>\n",
       "<td>0.0</td>\n",
       "<td>881.0</td>\n",
       "<td>3.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0123318</td>\n",
       "<td>11 / 892</td></tr>\n",
       "<tr><td>3.0</td>\n",
       "<td>3.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>9.0</td>\n",
       "<td>938.0</td>\n",
       "<td>0.0</td>\n",
       "<td>3.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0208768</td>\n",
       "<td>20 / 958</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>6.0</td>\n",
       "<td>5.0</td>\n",
       "<td>6.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>986.0</td>\n",
       "<td>1.0</td>\n",
       "<td>22.0</td>\n",
       "<td>0.0408560</td>\n",
       "<td>42 / 1,028</td></tr>\n",
       "<tr><td>7.0</td>\n",
       "<td>0.0</td>\n",
       "<td>4.0</td>\n",
       "<td>3.0</td>\n",
       "<td>0.0</td>\n",
       "<td>11.0</td>\n",
       "<td>5.0</td>\n",
       "<td>1.0</td>\n",
       "<td>942.0</td>\n",
       "<td>1.0</td>\n",
       "<td>0.0328542</td>\n",
       "<td>32 / 974</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>2.0</td>\n",
       "<td>1.0</td>\n",
       "<td>2.0</td>\n",
       "<td>6.0</td>\n",
       "<td>3.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>1.0</td>\n",
       "<td>993.0</td>\n",
       "<td>0.0158573</td>\n",
       "<td>16 / 1,009</td></tr>\n",
       "<tr><td>987.0</td>\n",
       "<td>1135.0</td>\n",
       "<td>1041.0</td>\n",
       "<td>1012.0</td>\n",
       "<td>980.0</td>\n",
       "<td>912.0</td>\n",
       "<td>952.0</td>\n",
       "<td>993.0</td>\n",
       "<td>960.0</td>\n",
       "<td>1028.0</td>\n",
       "<td>0.0184</td>\n",
       "<td>184 / 10,000</td></tr></table></div>"
      ],
      "text/plain": [
       "0    1     2     3     4    5    6    7    8    9     Error       Rate\n",
       "---  ----  ----  ----  ---  ---  ---  ---  ---  ----  ----------  ------------\n",
       "971  0     3     0     0    2    2    0    2    0     0.00918367  9 / 980\n",
       "4    1126  2     0     0    0    0    0    3    0     0.00792952  9 / 1,135\n",
       "1    2     1020  3     1    0    0    2    2    1     0.0116279   12 / 1,032\n",
       "0    0     2     993   0    6    0    3    5    1     0.0168317   17 / 1,010\n",
       "0    1     2     0     966  0    3    0    1    9     0.0162933   16 / 982\n",
       "1    0     0     6     0    881  3    0    0    1     0.0123318   11 / 892\n",
       "3    3     1     0     1    9    938  0    3    0     0.0208768   20 / 958\n",
       "0    1     6     5     6    0    1    986  1    22    0.040856    42 / 1,028\n",
       "7    0     4     3     0    11   5    1    942  1     0.0328542   32 / 974\n",
       "0    2     1     2     6    3    0    1    1    993   0.0158573   16 / 1,009\n",
       "987  1135  1041  1012  980  912  952  993  960  1028  0.0184      184 / 10,000"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Top-10 Hit Ratios: \n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b>k</b></td>\n",
       "<td><b>hit_ratio</b></td></tr>\n",
       "<tr><td>1</td>\n",
       "<td>0.9816</td></tr>\n",
       "<tr><td>2</td>\n",
       "<td>0.995</td></tr>\n",
       "<tr><td>3</td>\n",
       "<td>0.9986</td></tr>\n",
       "<tr><td>4</td>\n",
       "<td>0.999</td></tr>\n",
       "<tr><td>5</td>\n",
       "<td>0.9996</td></tr>\n",
       "<tr><td>6</td>\n",
       "<td>0.9997</td></tr>\n",
       "<tr><td>7</td>\n",
       "<td>0.9998</td></tr>\n",
       "<tr><td>8</td>\n",
       "<td>0.9998</td></tr>\n",
       "<tr><td>9</td>\n",
       "<td>0.9999000</td></tr>\n",
       "<tr><td>10</td>\n",
       "<td>1.0</td></tr></table></div>"
      ],
      "text/plain": [
       "k    hit_ratio\n",
       "---  -----------\n",
       "1    0.9816\n",
       "2    0.995\n",
       "3    0.9986\n",
       "4    0.999\n",
       "5    0.9996\n",
       "6    0.9997\n",
       "7    0.9998\n",
       "8    0.9998\n",
       "9    0.9999\n",
       "10   1"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "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>training_speed</b></td>\n",
       "<td><b>epochs</b></td>\n",
       "<td><b>iterations</b></td>\n",
       "<td><b>samples</b></td>\n",
       "<td><b>training_rmse</b></td>\n",
       "<td><b>training_logloss</b></td>\n",
       "<td><b>training_classification_error</b></td>\n",
       "<td><b>validation_rmse</b></td>\n",
       "<td><b>validation_logloss</b></td>\n",
       "<td><b>validation_classification_error</b></td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:52:44</td>\n",
       "<td> 0.000 sec</td>\n",
       "<td>None</td>\n",
       "<td>0.0</td>\n",
       "<td>0</td>\n",
       "<td>0.0</td>\n",
       "<td>nan</td>\n",
       "<td>nan</td>\n",
       "<td>nan</td>\n",
       "<td>nan</td>\n",
       "<td>nan</td>\n",
       "<td>nan</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:52:44</td>\n",
       "<td> 1.278 sec</td>\n",
       "<td>6872 obs/sec</td>\n",
       "<td>0.0170667</td>\n",
       "<td>1</td>\n",
       "<td>1024.0</td>\n",
       "<td>0.6147181</td>\n",
       "<td>1.1529294</td>\n",
       "<td>0.3878467</td>\n",
       "<td>0.6051395</td>\n",
       "<td>1.1116795</td>\n",
       "<td>0.377</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:52:54</td>\n",
       "<td>11.062 sec</td>\n",
       "<td>11805 obs/sec</td>\n",
       "<td>1.8090667</td>\n",
       "<td>106</td>\n",
       "<td>108544.0</td>\n",
       "<td>0.1552974</td>\n",
       "<td>0.1091712</td>\n",
       "<td>0.0285372</td>\n",
       "<td>0.1730931</td>\n",
       "<td>0.1352584</td>\n",
       "<td>0.0352</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:53:01</td>\n",
       "<td>18.577 sec</td>\n",
       "<td>11637 obs/sec</td>\n",
       "<td>3.0890667</td>\n",
       "<td>181</td>\n",
       "<td>185344.0</td>\n",
       "<td>0.1246167</td>\n",
       "<td>0.0618644</td>\n",
       "<td>0.0183596</td>\n",
       "<td>0.1434925</td>\n",
       "<td>0.0871844</td>\n",
       "<td>0.0239</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:53:09</td>\n",
       "<td>26.437 sec</td>\n",
       "<td>11490 obs/sec</td>\n",
       "<td>4.4032</td>\n",
       "<td>258</td>\n",
       "<td>264192.0</td>\n",
       "<td>0.1246678</td>\n",
       "<td>0.0744521</td>\n",
       "<td>0.0184594</td>\n",
       "<td>0.1465197</td>\n",
       "<td>0.1089135</td>\n",
       "<td>0.025</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:53:17</td>\n",
       "<td>34.419 sec</td>\n",
       "<td>11314 obs/sec</td>\n",
       "<td>5.7002667</td>\n",
       "<td>334</td>\n",
       "<td>342016.0</td>\n",
       "<td>0.1062530</td>\n",
       "<td>0.0479331</td>\n",
       "<td>0.0141688</td>\n",
       "<td>0.1341741</td>\n",
       "<td>0.0950406</td>\n",
       "<td>0.0209</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:53:25</td>\n",
       "<td>42.065 sec</td>\n",
       "<td>11151 obs/sec</td>\n",
       "<td>6.8778667</td>\n",
       "<td>403</td>\n",
       "<td>412672.0</td>\n",
       "<td>0.0938481</td>\n",
       "<td>0.0385263</td>\n",
       "<td>0.0105767</td>\n",
       "<td>0.1304027</td>\n",
       "<td>0.0858250</td>\n",
       "<td>0.0184</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:53:33</td>\n",
       "<td>50.665 sec</td>\n",
       "<td>11003 obs/sec</td>\n",
       "<td>8.2261333</td>\n",
       "<td>482</td>\n",
       "<td>493568.0</td>\n",
       "<td>0.0925869</td>\n",
       "<td>0.0449182</td>\n",
       "<td>0.0096787</td>\n",
       "<td>0.1340598</td>\n",
       "<td>0.1162025</td>\n",
       "<td>0.0202</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:53:41</td>\n",
       "<td>58.188 sec</td>\n",
       "<td>10888 obs/sec</td>\n",
       "<td>9.3696</td>\n",
       "<td>549</td>\n",
       "<td>562176.0</td>\n",
       "<td>0.0890942</td>\n",
       "<td>0.0356157</td>\n",
       "<td>0.0095789</td>\n",
       "<td>0.1383489</td>\n",
       "<td>0.1333587</td>\n",
       "<td>0.0215</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:53:48</td>\n",
       "<td> 1 min  5.805 sec</td>\n",
       "<td>10829 obs/sec</td>\n",
       "<td>10.5472</td>\n",
       "<td>618</td>\n",
       "<td>632832.0</td>\n",
       "<td>0.0605894</td>\n",
       "<td>0.0192707</td>\n",
       "<td>0.0045899</td>\n",
       "<td>0.1189931</td>\n",
       "<td>0.1052881</td>\n",
       "<td>0.0159</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:53:57</td>\n",
       "<td> 1 min 13.926 sec</td>\n",
       "<td>10785 obs/sec</td>\n",
       "<td>11.8272</td>\n",
       "<td>693</td>\n",
       "<td>709632.0</td>\n",
       "<td>0.0863284</td>\n",
       "<td>0.0428910</td>\n",
       "<td>0.0085811</td>\n",
       "<td>0.1295523</td>\n",
       "<td>0.1410438</td>\n",
       "<td>0.0189</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:54:04</td>\n",
       "<td> 1 min 21.612 sec</td>\n",
       "<td>10741 obs/sec</td>\n",
       "<td>13.0218667</td>\n",
       "<td>763</td>\n",
       "<td>781312.0</td>\n",
       "<td>0.0467808</td>\n",
       "<td>0.0107961</td>\n",
       "<td>0.0027939</td>\n",
       "<td>0.1263776</td>\n",
       "<td>0.1432967</td>\n",
       "<td>0.0175</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:54:12</td>\n",
       "<td> 1 min 29.209 sec</td>\n",
       "<td>10705 obs/sec</td>\n",
       "<td>14.1824</td>\n",
       "<td>831</td>\n",
       "<td>850944.0</td>\n",
       "<td>0.0551464</td>\n",
       "<td>0.0159995</td>\n",
       "<td>0.0036919</td>\n",
       "<td>0.1270667</td>\n",
       "<td>0.1650063</td>\n",
       "<td>0.0177</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:54:13</td>\n",
       "<td> 1 min 30.349 sec</td>\n",
       "<td>10688 obs/sec</td>\n",
       "<td>14.1824</td>\n",
       "<td>831</td>\n",
       "<td>850944.0</td>\n",
       "<td>0.0938481</td>\n",
       "<td>0.0385263</td>\n",
       "<td>0.0105767</td>\n",
       "<td>0.1304027</td>\n",
       "<td>0.0858250</td>\n",
       "<td>0.0184</td></tr></table></div>"
      ],
      "text/plain": [
       "    timestamp            duration          training_speed    epochs     iterations    samples    training_rmse    training_logloss    training_classification_error    validation_rmse    validation_logloss    validation_classification_error\n",
       "--  -------------------  ----------------  ----------------  ---------  ------------  ---------  ---------------  ------------------  -------------------------------  -----------------  --------------------  ---------------------------------\n",
       "    2017-05-09 04:52:44  0.000 sec                           0          0             0          nan              nan                 nan                              nan                nan                   nan\n",
       "    2017-05-09 04:52:44  1.278 sec         6872 obs/sec      0.0170667  1             1024       0.614718         1.15293             0.387847                         0.605139           1.11168               0.377\n",
       "    2017-05-09 04:52:54  11.062 sec        11805 obs/sec     1.80907    106           108544     0.155297         0.109171            0.0285372                        0.173093           0.135258              0.0352\n",
       "    2017-05-09 04:53:01  18.577 sec        11637 obs/sec     3.08907    181           185344     0.124617         0.0618644           0.0183596                        0.143493           0.0871844             0.0239\n",
       "    2017-05-09 04:53:09  26.437 sec        11490 obs/sec     4.4032     258           264192     0.124668         0.0744521           0.0184594                        0.14652            0.108913              0.025\n",
       "    2017-05-09 04:53:17  34.419 sec        11314 obs/sec     5.70027    334           342016     0.106253         0.0479331           0.0141688                        0.134174           0.0950406             0.0209\n",
       "    2017-05-09 04:53:25  42.065 sec        11151 obs/sec     6.87787    403           412672     0.0938481        0.0385263           0.0105767                        0.130403           0.085825              0.0184\n",
       "    2017-05-09 04:53:33  50.665 sec        11003 obs/sec     8.22613    482           493568     0.0925869        0.0449182           0.00967871                       0.13406            0.116202              0.0202\n",
       "    2017-05-09 04:53:41  58.188 sec        10888 obs/sec     9.3696     549           562176     0.0890942        0.0356157           0.00957893                       0.138349           0.133359              0.0215\n",
       "    2017-05-09 04:53:48  1 min  5.805 sec  10829 obs/sec     10.5472    618           632832     0.0605894        0.0192707           0.0045899                        0.118993           0.105288              0.0159\n",
       "    2017-05-09 04:53:57  1 min 13.926 sec  10785 obs/sec     11.8272    693           709632     0.0863284        0.042891            0.00858112                       0.129552           0.141044              0.0189\n",
       "    2017-05-09 04:54:04  1 min 21.612 sec  10741 obs/sec     13.0219    763           781312     0.0467808        0.0107961           0.00279385                       0.126378           0.143297              0.0175\n",
       "    2017-05-09 04:54:12  1 min 29.209 sec  10705 obs/sec     14.1824    831           850944     0.0551464        0.0159995           0.00369188                       0.127067           0.165006              0.0177\n",
       "    2017-05-09 04:54:13  1 min 30.349 sec  10688 obs/sec     14.1824    831           850944     0.0938481        0.0385263           0.0105767                        0.130403           0.085825              0.0184"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model_mnist_lenet_tf.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Build Deep Water Caffe Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "deepwater Model Build progress: |█████████████████████████████████████████| 100%\n"
     ]
    }
   ],
   "source": [
    "model_mnist_lenet_caffe = H2ODeepWaterEstimator(epochs=80, network=\"lenet\", backend=\"caffe\")\n",
    "model_mnist_lenet_caffe.train(x=[\"uri\"], y=\"label\",\n",
    "                           training_frame=mnist_training,\n",
    "                           validation_frame=mnist_testing,\n",
    "                           model_id=\"model_mnist_lenet_caffe\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model Details\n",
      "=============\n",
      "H2ODeepWaterEstimator :  Deep Water\n",
      "Model Key:  model_mnist_lenet_caffe\n",
      "\n",
      "\n",
      "ModelMetricsMultinomial: deepwater\n",
      "** Reported on train data. **\n",
      "\n",
      "MSE: 0.8095571993929722\n",
      "RMSE: 0.8997539660334776\n",
      "LogLoss: 2.301585012019263\n",
      "Mean Per-Class Error: 0.9\n",
      "Confusion Matrix: vertical: actual; across: predicted\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b>0</b></td>\n",
       "<td><b>1</b></td>\n",
       "<td><b>2</b></td>\n",
       "<td><b>3</b></td>\n",
       "<td><b>4</b></td>\n",
       "<td><b>5</b></td>\n",
       "<td><b>6</b></td>\n",
       "<td><b>7</b></td>\n",
       "<td><b>8</b></td>\n",
       "<td><b>9</b></td>\n",
       "<td><b>Error</b></td>\n",
       "<td><b>Rate</b></td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>984.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>984 / 984</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>1120.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0 / 1,120</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>1020.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>1,020 / 1,020</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>1034.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>1,034 / 1,034</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>997.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>997 / 997</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>897.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>897 / 897</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>990.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>990 / 990</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>1047.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>1,047 / 1,047</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>995.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>995 / 995</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>985.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>985 / 985</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>10069.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.8887675</td>\n",
       "<td>8,949 / 10,069</td></tr></table></div>"
      ],
      "text/plain": [
       "0    1      2    3    4    5    6    7    8    9    Error     Rate\n",
       "---  -----  ---  ---  ---  ---  ---  ---  ---  ---  --------  --------------\n",
       "0    984    0    0    0    0    0    0    0    0    1         984 / 984\n",
       "0    1120   0    0    0    0    0    0    0    0    0         0 / 1,120\n",
       "0    1020   0    0    0    0    0    0    0    0    1         1,020 / 1,020\n",
       "0    1034   0    0    0    0    0    0    0    0    1         1,034 / 1,034\n",
       "0    997    0    0    0    0    0    0    0    0    1         997 / 997\n",
       "0    897    0    0    0    0    0    0    0    0    1         897 / 897\n",
       "0    990    0    0    0    0    0    0    0    0    1         990 / 990\n",
       "0    1047   0    0    0    0    0    0    0    0    1         1,047 / 1,047\n",
       "0    995    0    0    0    0    0    0    0    0    1         995 / 995\n",
       "0    985    0    0    0    0    0    0    0    0    1         985 / 985\n",
       "0    10069  0    0    0    0    0    0    0    0    0.888768  8,949 / 10,069"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Top-10 Hit Ratios: \n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b>k</b></td>\n",
       "<td><b>hit_ratio</b></td></tr>\n",
       "<tr><td>1</td>\n",
       "<td>0.1112325</td></tr>\n",
       "<tr><td>2</td>\n",
       "<td>0.2139239</td></tr>\n",
       "<tr><td>3</td>\n",
       "<td>0.3129407</td></tr>\n",
       "<tr><td>4</td>\n",
       "<td>0.4169232</td></tr>\n",
       "<tr><td>5</td>\n",
       "<td>0.5147482</td></tr>\n",
       "<tr><td>6</td>\n",
       "<td>0.6124739</td></tr>\n",
       "<tr><td>7</td>\n",
       "<td>0.711292</td></tr>\n",
       "<tr><td>8</td>\n",
       "<td>0.8125930</td></tr>\n",
       "<tr><td>9</td>\n",
       "<td>0.9109146</td></tr>\n",
       "<tr><td>10</td>\n",
       "<td>0.9999999</td></tr></table></div>"
      ],
      "text/plain": [
       "k    hit_ratio\n",
       "---  -----------\n",
       "1    0.111232\n",
       "2    0.213924\n",
       "3    0.312941\n",
       "4    0.416923\n",
       "5    0.514748\n",
       "6    0.612474\n",
       "7    0.711292\n",
       "8    0.812593\n",
       "9    0.910915\n",
       "10   1"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "ModelMetricsMultinomial: deepwater\n",
      "** Reported on validation data. **\n",
      "\n",
      "MSE: 0.8095325554508053\n",
      "RMSE: 0.899740271106504\n",
      "LogLoss: 2.3014509464429227\n",
      "Mean Per-Class Error: 0.9\n",
      "Confusion Matrix: vertical: actual; across: predicted\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b>0</b></td>\n",
       "<td><b>1</b></td>\n",
       "<td><b>2</b></td>\n",
       "<td><b>3</b></td>\n",
       "<td><b>4</b></td>\n",
       "<td><b>5</b></td>\n",
       "<td><b>6</b></td>\n",
       "<td><b>7</b></td>\n",
       "<td><b>8</b></td>\n",
       "<td><b>9</b></td>\n",
       "<td><b>Error</b></td>\n",
       "<td><b>Rate</b></td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>980.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>980 / 980</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>1135.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0 / 1,135</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>1032.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>1,032 / 1,032</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>1010.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>1,010 / 1,010</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>982.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>982 / 982</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>892.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>892 / 892</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>958.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>958 / 958</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>1028.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>1,028 / 1,028</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>974.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>974 / 974</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>1009.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>1.0</td>\n",
       "<td>1,009 / 1,009</td></tr>\n",
       "<tr><td>0.0</td>\n",
       "<td>10000.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.0</td>\n",
       "<td>0.8865</td>\n",
       "<td>8,865 / 10,000</td></tr></table></div>"
      ],
      "text/plain": [
       "0    1      2    3    4    5    6    7    8    9    Error    Rate\n",
       "---  -----  ---  ---  ---  ---  ---  ---  ---  ---  -------  --------------\n",
       "0    980    0    0    0    0    0    0    0    0    1        980 / 980\n",
       "0    1135   0    0    0    0    0    0    0    0    0        0 / 1,135\n",
       "0    1032   0    0    0    0    0    0    0    0    1        1,032 / 1,032\n",
       "0    1010   0    0    0    0    0    0    0    0    1        1,010 / 1,010\n",
       "0    982    0    0    0    0    0    0    0    0    1        982 / 982\n",
       "0    892    0    0    0    0    0    0    0    0    1        892 / 892\n",
       "0    958    0    0    0    0    0    0    0    0    1        958 / 958\n",
       "0    1028   0    0    0    0    0    0    0    0    1        1,028 / 1,028\n",
       "0    974    0    0    0    0    0    0    0    0    1        974 / 974\n",
       "0    1009   0    0    0    0    0    0    0    0    1        1,009 / 1,009\n",
       "0    10000  0    0    0    0    0    0    0    0    0.8865   8,865 / 10,000"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Top-10 Hit Ratios: \n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b>k</b></td>\n",
       "<td><b>hit_ratio</b></td></tr>\n",
       "<tr><td>1</td>\n",
       "<td>0.1135</td></tr>\n",
       "<tr><td>2</td>\n",
       "<td>0.2145000</td></tr>\n",
       "<tr><td>3</td>\n",
       "<td>0.3127</td></tr>\n",
       "<tr><td>4</td>\n",
       "<td>0.4155</td></tr>\n",
       "<tr><td>5</td>\n",
       "<td>0.5164</td></tr>\n",
       "<tr><td>6</td>\n",
       "<td>0.6144000</td></tr>\n",
       "<tr><td>7</td>\n",
       "<td>0.7118</td></tr>\n",
       "<tr><td>8</td>\n",
       "<td>0.815</td></tr>\n",
       "<tr><td>9</td>\n",
       "<td>0.9108</td></tr>\n",
       "<tr><td>10</td>\n",
       "<td>1.0</td></tr></table></div>"
      ],
      "text/plain": [
       "k    hit_ratio\n",
       "---  -----------\n",
       "1    0.1135\n",
       "2    0.2145\n",
       "3    0.3127\n",
       "4    0.4155\n",
       "5    0.5164\n",
       "6    0.6144\n",
       "7    0.7118\n",
       "8    0.815\n",
       "9    0.9108\n",
       "10   1"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "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>training_speed</b></td>\n",
       "<td><b>epochs</b></td>\n",
       "<td><b>iterations</b></td>\n",
       "<td><b>samples</b></td>\n",
       "<td><b>training_rmse</b></td>\n",
       "<td><b>training_logloss</b></td>\n",
       "<td><b>training_classification_error</b></td>\n",
       "<td><b>validation_rmse</b></td>\n",
       "<td><b>validation_logloss</b></td>\n",
       "<td><b>validation_classification_error</b></td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:54:16</td>\n",
       "<td> 0.000 sec</td>\n",
       "<td>None</td>\n",
       "<td>0.0</td>\n",
       "<td>0</td>\n",
       "<td>0.0</td>\n",
       "<td>nan</td>\n",
       "<td>nan</td>\n",
       "<td>nan</td>\n",
       "<td>nan</td>\n",
       "<td>nan</td>\n",
       "<td>nan</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:54:16</td>\n",
       "<td> 4.941 sec</td>\n",
       "<td>3518 obs/sec</td>\n",
       "<td>0.0170667</td>\n",
       "<td>1</td>\n",
       "<td>1024.0</td>\n",
       "<td>0.8668129</td>\n",
       "<td>24.1941831</td>\n",
       "<td>0.7544940</td>\n",
       "<td>0.8608721</td>\n",
       "<td>23.8999989</td>\n",
       "<td>0.744</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:54:49</td>\n",
       "<td>38.188 sec</td>\n",
       "<td>3889 obs/sec</td>\n",
       "<td>1.9456</td>\n",
       "<td>114</td>\n",
       "<td>116736.0</td>\n",
       "<td>0.8996859</td>\n",
       "<td>2.3044751</td>\n",
       "<td>0.8887675</td>\n",
       "<td>0.8995699</td>\n",
       "<td>2.3033484</td>\n",
       "<td>0.8865</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:55:25</td>\n",
       "<td> 1 min 13.475 sec</td>\n",
       "<td>3679 obs/sec</td>\n",
       "<td>3.7888</td>\n",
       "<td>222</td>\n",
       "<td>227328.0</td>\n",
       "<td>0.8998119</td>\n",
       "<td>2.3063774</td>\n",
       "<td>0.8960175</td>\n",
       "<td>0.8999011</td>\n",
       "<td>2.3072931</td>\n",
       "<td>0.8972</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:56:00</td>\n",
       "<td> 1 min 49.227 sec</td>\n",
       "<td>3626 obs/sec</td>\n",
       "<td>5.6661333</td>\n",
       "<td>332</td>\n",
       "<td>339968.0</td>\n",
       "<td>0.8998142</td>\n",
       "<td>2.3035552</td>\n",
       "<td>0.9021750</td>\n",
       "<td>0.8997164</td>\n",
       "<td>2.3025943</td>\n",
       "<td>0.8991</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:56:38</td>\n",
       "<td> 2 min 26.886 sec</td>\n",
       "<td>3597 obs/sec</td>\n",
       "<td>7.6629333</td>\n",
       "<td>449</td>\n",
       "<td>459776.0</td>\n",
       "<td>0.8997377</td>\n",
       "<td>2.3031960</td>\n",
       "<td>0.9011818</td>\n",
       "<td>0.8997106</td>\n",
       "<td>2.3029205</td>\n",
       "<td>0.9026</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:57:14</td>\n",
       "<td> 3 min  3.275 sec</td>\n",
       "<td>3578 obs/sec</td>\n",
       "<td>9.5744</td>\n",
       "<td>561</td>\n",
       "<td>574464.0</td>\n",
       "<td>0.8997930</td>\n",
       "<td>2.3031519</td>\n",
       "<td>0.8887675</td>\n",
       "<td>0.8997469</td>\n",
       "<td>2.3027048</td>\n",
       "<td>0.8865</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:57:51</td>\n",
       "<td> 3 min 40.322 sec</td>\n",
       "<td>3560 obs/sec</td>\n",
       "<td>11.5029333</td>\n",
       "<td>674</td>\n",
       "<td>690176.0</td>\n",
       "<td>0.8996081</td>\n",
       "<td>2.3037517</td>\n",
       "<td>0.8887675</td>\n",
       "<td>0.8995709</td>\n",
       "<td>2.3034134</td>\n",
       "<td>0.8865</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:58:29</td>\n",
       "<td> 4 min 17.371 sec</td>\n",
       "<td>3557 obs/sec</td>\n",
       "<td>13.4826667</td>\n",
       "<td>790</td>\n",
       "<td>808960.0</td>\n",
       "<td>0.8997540</td>\n",
       "<td>2.3015850</td>\n",
       "<td>0.8887675</td>\n",
       "<td>0.8997403</td>\n",
       "<td>2.3014509</td>\n",
       "<td>0.8865</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:59:04</td>\n",
       "<td> 4 min 52.820 sec</td>\n",
       "<td>3567 obs/sec</td>\n",
       "<td>15.4112</td>\n",
       "<td>903</td>\n",
       "<td>924672.0</td>\n",
       "<td>0.8999979</td>\n",
       "<td>2.3047100</td>\n",
       "<td>0.9011818</td>\n",
       "<td>0.9000687</td>\n",
       "<td>2.3054582</td>\n",
       "<td>0.9026</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 04:59:40</td>\n",
       "<td> 5 min 29.482 sec</td>\n",
       "<td>3569 obs/sec</td>\n",
       "<td>17.3738667</td>\n",
       "<td>1018</td>\n",
       "<td>1042432.0</td>\n",
       "<td>0.8998989</td>\n",
       "<td>2.3048571</td>\n",
       "<td>0.8960175</td>\n",
       "<td>0.8998659</td>\n",
       "<td>2.3045051</td>\n",
       "<td>0.8972</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 05:00:19</td>\n",
       "<td> 6 min  7.478 sec</td>\n",
       "<td>3554 obs/sec</td>\n",
       "<td>19.3365333</td>\n",
       "<td>1133</td>\n",
       "<td>1160192.0</td>\n",
       "<td>0.8995578</td>\n",
       "<td>2.3032201</td>\n",
       "<td>0.8887675</td>\n",
       "<td>0.8995016</td>\n",
       "<td>2.3026755</td>\n",
       "<td>0.8865</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 05:00:56</td>\n",
       "<td> 6 min 44.710 sec</td>\n",
       "<td>3546 obs/sec</td>\n",
       "<td>21.2821333</td>\n",
       "<td>1247</td>\n",
       "<td>1276928.0</td>\n",
       "<td>0.8998191</td>\n",
       "<td>2.3033039</td>\n",
       "<td>0.8973086</td>\n",
       "<td>0.8998263</td>\n",
       "<td>2.3033491</td>\n",
       "<td>0.899</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 05:01:32</td>\n",
       "<td> 7 min 20.805 sec</td>\n",
       "<td>3541 obs/sec</td>\n",
       "<td>23.1594667</td>\n",
       "<td>1357</td>\n",
       "<td>1389568.0</td>\n",
       "<td>0.8996398</td>\n",
       "<td>2.3031084</td>\n",
       "<td>0.8887675</td>\n",
       "<td>0.8995644</td>\n",
       "<td>2.3023834</td>\n",
       "<td>0.8865</td></tr>\n",
       "<tr><td></td>\n",
       "<td>2017-05-09 05:01:36</td>\n",
       "<td> 7 min 27.068 sec</td>\n",
       "<td>3541 obs/sec</td>\n",
       "<td>23.1594667</td>\n",
       "<td>1357</td>\n",
       "<td>1389568.0</td>\n",
       "<td>0.8997540</td>\n",
       "<td>2.3015850</td>\n",
       "<td>0.8887675</td>\n",
       "<td>0.8997403</td>\n",
       "<td>2.3014509</td>\n",
       "<td>0.8865</td></tr></table></div>"
      ],
      "text/plain": [
       "    timestamp            duration          training_speed    epochs     iterations    samples      training_rmse    training_logloss    training_classification_error    validation_rmse    validation_logloss    validation_classification_error\n",
       "--  -------------------  ----------------  ----------------  ---------  ------------  -----------  ---------------  ------------------  -------------------------------  -----------------  --------------------  ---------------------------------\n",
       "    2017-05-09 04:54:16  0.000 sec                           0          0             0            nan              nan                 nan                              nan                nan                   nan\n",
       "    2017-05-09 04:54:16  4.941 sec         3518 obs/sec      0.0170667  1             1024         0.866813         24.1942             0.754494                         0.860872           23.9                  0.744\n",
       "    2017-05-09 04:54:49  38.188 sec        3889 obs/sec      1.9456     114           116736       0.899686         2.30448             0.888768                         0.89957            2.30335               0.8865\n",
       "    2017-05-09 04:55:25  1 min 13.475 sec  3679 obs/sec      3.7888     222           227328       0.899812         2.30638             0.896017                         0.899901           2.30729               0.8972\n",
       "    2017-05-09 04:56:00  1 min 49.227 sec  3626 obs/sec      5.66613    332           339968       0.899814         2.30356             0.902175                         0.899716           2.30259               0.8991\n",
       "    2017-05-09 04:56:38  2 min 26.886 sec  3597 obs/sec      7.66293    449           459776       0.899738         2.3032              0.901182                         0.899711           2.30292               0.9026\n",
       "    2017-05-09 04:57:14  3 min  3.275 sec  3578 obs/sec      9.5744     561           574464       0.899793         2.30315             0.888768                         0.899747           2.3027                0.8865\n",
       "    2017-05-09 04:57:51  3 min 40.322 sec  3560 obs/sec      11.5029    674           690176       0.899608         2.30375             0.888768                         0.899571           2.30341               0.8865\n",
       "    2017-05-09 04:58:29  4 min 17.371 sec  3557 obs/sec      13.4827    790           808960       0.899754         2.30159             0.888768                         0.89974            2.30145               0.8865\n",
       "    2017-05-09 04:59:04  4 min 52.820 sec  3567 obs/sec      15.4112    903           924672       0.899998         2.30471             0.901182                         0.900069           2.30546               0.9026\n",
       "    2017-05-09 04:59:40  5 min 29.482 sec  3569 obs/sec      17.3739    1018          1.04243e+06  0.899899         2.30486             0.896017                         0.899866           2.30451               0.8972\n",
       "    2017-05-09 05:00:19  6 min  7.478 sec  3554 obs/sec      19.3365    1133          1.16019e+06  0.899558         2.30322             0.888768                         0.899502           2.30268               0.8865\n",
       "    2017-05-09 05:00:56  6 min 44.710 sec  3546 obs/sec      21.2821    1247          1.27693e+06  0.899819         2.3033              0.897309                         0.899826           2.30335               0.899\n",
       "    2017-05-09 05:01:32  7 min 20.805 sec  3541 obs/sec      23.1595    1357          1.38957e+06  0.89964          2.30311             0.888768                         0.899564           2.30238               0.8865\n",
       "    2017-05-09 05:01:36  7 min 27.068 sec  3541 obs/sec      23.1595    1357          1.38957e+06  0.899754         2.30159             0.888768                         0.89974            2.30145               0.8865"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model_mnist_lenet_caffe.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Save H2O Models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'/gtc-2017/models/model_mnist_lenet_caffe'"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "h2o.save_model(model=model_mnist_lenet_mx, path=project_path+\"/models/\", force=True)\n",
    "h2o.save_model(model=model_mnist_lenet_tf, path=project_path+\"/models/\", force=True)\n",
    "h2o.save_model(model=model_mnist_lenet_caffe, path=project_path+\"/models/\", force=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Export MOJOs for Deployment"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'/gtc-2017/models/model_mnist_lenet_caffe.zip'"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_mnist_lenet_mx.download_mojo(path=project_path+\"/models/\", get_genmodel_jar=True)\n",
    "model_mnist_lenet_tf.download_mojo(path=project_path+\"/models/\", get_genmodel_jar=True)\n",
    "model_mnist_lenet_caffe.download_mojo(path=project_path+\"/models/\", get_genmodel_jar=True)"
   ]
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "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": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}