jet_exploration.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.layers import Dense, Input, Conv2D, Conv2DTranspose, Lambda, Flatten, Reshape, MaxPooling2D\n",
    "from keras.models import Model\n",
    "from keras.losses import mean_squared_error, binary_crossentropy, kullback_leibler_divergence\n",
    "from keras.callbacks import EarlyStopping, TerminateOnNaN, ReduceLROnPlateau\n",
    "from keras import backend as K\n",
    "import tensorflow as tf\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import os, scipy\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "from phenom_read_in_data import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Getthe data\n",
    "df = csv_to_df('ttbarZ/')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>event_ID</th>\n",
       "      <th>process_ID</th>\n",
       "      <th>event_weight</th>\n",
       "      <th>MET</th>\n",
       "      <th>MET_Phi</th>\n",
       "      <th>obj1</th>\n",
       "      <th>E1</th>\n",
       "      <th>pt1</th>\n",
       "      <th>eta1</th>\n",
       "      <th>phi1</th>\n",
       "      <th>...</th>\n",
       "      <th>obj12</th>\n",
       "      <th>E12</th>\n",
       "      <th>pt12</th>\n",
       "      <th>eta12</th>\n",
       "      <th>phi12</th>\n",
       "      <th>obj13</th>\n",
       "      <th>E13</th>\n",
       "      <th>pt13</th>\n",
       "      <th>eta13</th>\n",
       "      <th>phi13</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>21</td>\n",
       "      <td>ttbarZ</td>\n",
       "      <td>1</td>\n",
       "      <td>43924.6</td>\n",
       "      <td>-0.38825</td>\n",
       "      <td>j</td>\n",
       "      <td>165476</td>\n",
       "      <td>165097</td>\n",
       "      <td>-0.0171131</td>\n",
       "      <td>-1.54448</td>\n",
       "      <td>...</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>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>21</td>\n",
       "      <td>ttbarZ</td>\n",
       "      <td>1</td>\n",
       "      <td>45289</td>\n",
       "      <td>-3.09984</td>\n",
       "      <td>j</td>\n",
       "      <td>105746</td>\n",
       "      <td>95333.9</td>\n",
       "      <td>0.449145</td>\n",
       "      <td>1.79273</td>\n",
       "      <td>...</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>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>21</td>\n",
       "      <td>ttbarZ</td>\n",
       "      <td>1</td>\n",
       "      <td>16530.4</td>\n",
       "      <td>-3.09183</td>\n",
       "      <td>b</td>\n",
       "      <td>144211</td>\n",
       "      <td>137971</td>\n",
       "      <td>0.100321</td>\n",
       "      <td>-0.0772316</td>\n",
       "      <td>...</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>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>21</td>\n",
       "      <td>ttbarZ</td>\n",
       "      <td>1</td>\n",
       "      <td>18362.9</td>\n",
       "      <td>1.14058</td>\n",
       "      <td>b</td>\n",
       "      <td>370358</td>\n",
       "      <td>120835</td>\n",
       "      <td>1.78374</td>\n",
       "      <td>-0.757091</td>\n",
       "      <td>...</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>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>21</td>\n",
       "      <td>ttbarZ</td>\n",
       "      <td>1</td>\n",
       "      <td>232797</td>\n",
       "      <td>-2.74669</td>\n",
       "      <td>j</td>\n",
       "      <td>436613</td>\n",
       "      <td>432317</td>\n",
       "      <td>-0.110325</td>\n",
       "      <td>0.656242</td>\n",
       "      <td>...</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>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5869</td>\n",
       "      <td>21</td>\n",
       "      <td>ttbarZ</td>\n",
       "      <td>1</td>\n",
       "      <td>17569.3</td>\n",
       "      <td>3.08905</td>\n",
       "      <td>j</td>\n",
       "      <td>223425</td>\n",
       "      <td>161759</td>\n",
       "      <td>-0.840404</td>\n",
       "      <td>2.52008</td>\n",
       "      <td>...</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>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5870</td>\n",
       "      <td>21</td>\n",
       "      <td>ttbarZ</td>\n",
       "      <td>1</td>\n",
       "      <td>70402.4</td>\n",
       "      <td>-1.12592</td>\n",
       "      <td>b</td>\n",
       "      <td>340938</td>\n",
       "      <td>86263.4</td>\n",
       "      <td>-2.05015</td>\n",
       "      <td>-0.018614</td>\n",
       "      <td>...</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>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5871</td>\n",
       "      <td>21</td>\n",
       "      <td>ttbarZ</td>\n",
       "      <td>1</td>\n",
       "      <td>28553.9</td>\n",
       "      <td>0.862966</td>\n",
       "      <td>j</td>\n",
       "      <td>116154</td>\n",
       "      <td>114311</td>\n",
       "      <td>0.0884322</td>\n",
       "      <td>-2.74727</td>\n",
       "      <td>...</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>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5872</td>\n",
       "      <td>21</td>\n",
       "      <td>ttbarZ</td>\n",
       "      <td>1</td>\n",
       "      <td>183633</td>\n",
       "      <td>1.98713</td>\n",
       "      <td>j</td>\n",
       "      <td>350312</td>\n",
       "      <td>178245</td>\n",
       "      <td>1.29249</td>\n",
       "      <td>0.477881</td>\n",
       "      <td>...</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>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5873</td>\n",
       "      <td>21</td>\n",
       "      <td>ttbarZ</td>\n",
       "      <td>1</td>\n",
       "      <td>34380</td>\n",
       "      <td>0.550302</td>\n",
       "      <td>j</td>\n",
       "      <td>402447</td>\n",
       "      <td>97361.7</td>\n",
       "      <td>2.09636</td>\n",
       "      <td>-2.32422</td>\n",
       "      <td>...</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>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5874 rows × 70 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     event_ID process_ID event_weight      MET   MET_Phi obj1      E1  \\\n",
       "0          21     ttbarZ            1  43924.6  -0.38825    j  165476   \n",
       "1          21     ttbarZ            1    45289  -3.09984    j  105746   \n",
       "2          21     ttbarZ            1  16530.4  -3.09183    b  144211   \n",
       "3          21     ttbarZ            1  18362.9   1.14058    b  370358   \n",
       "4          21     ttbarZ            1   232797  -2.74669    j  436613   \n",
       "...       ...        ...          ...      ...       ...  ...     ...   \n",
       "5869       21     ttbarZ            1  17569.3   3.08905    j  223425   \n",
       "5870       21     ttbarZ            1  70402.4  -1.12592    b  340938   \n",
       "5871       21     ttbarZ            1  28553.9  0.862966    j  116154   \n",
       "5872       21     ttbarZ            1   183633   1.98713    j  350312   \n",
       "5873       21     ttbarZ            1    34380  0.550302    j  402447   \n",
       "\n",
       "          pt1        eta1        phi1  ... obj12  E12 pt12 eta12 phi12 obj13  \\\n",
       "0      165097  -0.0171131    -1.54448  ...   NaN  NaN  NaN   NaN   NaN   NaN   \n",
       "1     95333.9    0.449145     1.79273  ...   NaN  NaN  NaN   NaN   NaN   NaN   \n",
       "2      137971    0.100321  -0.0772316  ...   NaN  NaN  NaN   NaN   NaN   NaN   \n",
       "3      120835     1.78374   -0.757091  ...   NaN  NaN  NaN   NaN   NaN   NaN   \n",
       "4      432317   -0.110325    0.656242  ...   NaN  NaN  NaN   NaN   NaN   NaN   \n",
       "...       ...         ...         ...  ...   ...  ...  ...   ...   ...   ...   \n",
       "5869   161759   -0.840404     2.52008  ...   NaN  NaN  NaN   NaN   NaN   NaN   \n",
       "5870  86263.4    -2.05015   -0.018614  ...   NaN  NaN  NaN   NaN   NaN   NaN   \n",
       "5871   114311   0.0884322    -2.74727  ...   NaN  NaN  NaN   NaN   NaN   NaN   \n",
       "5872   178245     1.29249    0.477881  ...   NaN  NaN  NaN   NaN   NaN   NaN   \n",
       "5873  97361.7     2.09636    -2.32422  ...   NaN  NaN  NaN   NaN   NaN   NaN   \n",
       "\n",
       "      E13 pt13 eta13 phi13  \n",
       "0     NaN  NaN   NaN   NaN  \n",
       "1     NaN  NaN   NaN   NaN  \n",
       "2     NaN  NaN   NaN   NaN  \n",
       "3     NaN  NaN   NaN   NaN  \n",
       "4     NaN  NaN   NaN   NaN  \n",
       "...   ...  ...   ...   ...  \n",
       "5869  NaN  NaN   NaN   NaN  \n",
       "5870  NaN  NaN   NaN   NaN  \n",
       "5871  NaN  NaN   NaN   NaN  \n",
       "5872  NaN  NaN   NaN   NaN  \n",
       "5873  NaN  NaN   NaN   NaN  \n",
       "\n",
       "[5874 rows x 70 columns]"
      ]
     },
     "execution_count": 77,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [],
   "source": [
    "#If the number of max objects is prime, another column needs to be added to mitigate "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [],
   "source": [
    "df['obj14']= -999\n",
    "df['E14'] = -999\n",
    "df['pt14'] = -999\n",
    "df['eta14'] = -999\n",
    "df['phi14'] = -999"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = df.fillna(-999)\n",
    "variables = [entry for entry in df.columns if entry[0] == 'E'] + [entry for entry in df.columns if entry[0:2] == 'pt'] + [entry for entry in df.columns if entry[0:2] == 'et'] + [entry for entry in df.columns if entry[0:2] == 'ph']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [],
   "source": [
    "objects = []\n",
    "for i in df.columns:\n",
    "    if i[:3] == 'obj':\n",
    "        objects.append(i)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [],
   "source": [
    "j_objects1 = df.loc[df['obj1'] == 'j', df.columns[6:10]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>E1</th>\n",
       "      <th>pt1</th>\n",
       "      <th>eta1</th>\n",
       "      <th>phi1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>165476</td>\n",
       "      <td>165097</td>\n",
       "      <td>-0.0171131</td>\n",
       "      <td>-1.54448</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>105746</td>\n",
       "      <td>95333.9</td>\n",
       "      <td>0.449145</td>\n",
       "      <td>1.79273</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>436613</td>\n",
       "      <td>432317</td>\n",
       "      <td>-0.110325</td>\n",
       "      <td>0.656242</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>265871</td>\n",
       "      <td>210909</td>\n",
       "      <td>-0.671705</td>\n",
       "      <td>-2.92575</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>296867</td>\n",
       "      <td>102488</td>\n",
       "      <td>-1.72479</td>\n",
       "      <td>1.81765</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5868</td>\n",
       "      <td>227671</td>\n",
       "      <td>149490</td>\n",
       "      <td>-0.959438</td>\n",
       "      <td>-2.704</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5869</td>\n",
       "      <td>223425</td>\n",
       "      <td>161759</td>\n",
       "      <td>-0.840404</td>\n",
       "      <td>2.52008</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5871</td>\n",
       "      <td>116154</td>\n",
       "      <td>114311</td>\n",
       "      <td>0.0884322</td>\n",
       "      <td>-2.74727</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5872</td>\n",
       "      <td>350312</td>\n",
       "      <td>178245</td>\n",
       "      <td>1.29249</td>\n",
       "      <td>0.477881</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5873</td>\n",
       "      <td>402447</td>\n",
       "      <td>97361.7</td>\n",
       "      <td>2.09636</td>\n",
       "      <td>-2.32422</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>3967 rows × 4 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          E1      pt1        eta1      phi1\n",
       "0     165476   165097  -0.0171131  -1.54448\n",
       "1     105746  95333.9    0.449145   1.79273\n",
       "4     436613   432317   -0.110325  0.656242\n",
       "5     265871   210909   -0.671705  -2.92575\n",
       "6     296867   102488    -1.72479   1.81765\n",
       "...      ...      ...         ...       ...\n",
       "5868  227671   149490   -0.959438    -2.704\n",
       "5869  223425   161759   -0.840404   2.52008\n",
       "5871  116154   114311   0.0884322  -2.74727\n",
       "5872  350312   178245     1.29249  0.477881\n",
       "5873  402447  97361.7     2.09636  -2.32422\n",
       "\n",
       "[3967 rows x 4 columns]"
      ]
     },
     "execution_count": 83,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "j_objects1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [],
   "source": [
    "j_objects2 = df.loc[df['obj2'] == 'j', df.columns[11:15]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/joe/anaconda3/lib/python3.7/site-packages/pandas/core/ops/__init__.py:1115: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  result = method(y)\n"
     ]
    }
   ],
   "source": [
    "j_objects2 = df.loc[df['obj2'] == 'j', df.columns[11:15]]\n",
    "j_objects3 = df.loc[df['obj3'] == 'j', df.columns[16:20]]\n",
    "j_objects4 = df.loc[df['obj4'] == 'j', df.columns[21:25]]\n",
    "j_objects5 = df.loc[df['obj5'] == 'j', df.columns[26:30]]\n",
    "j_objects6 = df.loc[df['obj6'] == 'j', df.columns[31:35]]\n",
    "j_objects7 = df.loc[df['obj7'] == 'j', df.columns[36:40]]\n",
    "j_objects8 = df.loc[df['obj8'] == 'j', df.columns[41:45]]\n",
    "j_objects9 = df.loc[df['obj9'] == 'j', df.columns[46:50]]\n",
    "j_objects10 = df.loc[df['obj10'] == 'j', df.columns[51:55]]\n",
    "j_objects11 = df.loc[df['obj11'] == 'j', df.columns[56:60]]\n",
    "j_objects12 = df.loc[df['obj12'] == 'j', df.columns[61:65]]\n",
    "j_objects13 = df.loc[df['obj13'] == 'j', df.columns[66:70]]\n",
    "j_objects14 = df.loc[df['obj14'] == 'j', df.columns[71:75]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [],
   "source": [
    "j_objs = [j_objects1, j_objects2, j_objects3, j_objects4, j_objects5, j_objects6, j_objects7, j_objects8, \n",
    "         j_objects9, j_objects10, j_objects11, j_objects12, j_objects13, j_objects14]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = pd.concat(j_objs, sort=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>E1</th>\n",
       "      <th>pt1</th>\n",
       "      <th>eta1</th>\n",
       "      <th>phi1</th>\n",
       "      <th>E2</th>\n",
       "      <th>pt2</th>\n",
       "      <th>eta2</th>\n",
       "      <th>phi2</th>\n",
       "      <th>E3</th>\n",
       "      <th>pt3</th>\n",
       "      <th>...</th>\n",
       "      <th>eta12</th>\n",
       "      <th>phi12</th>\n",
       "      <th>E13</th>\n",
       "      <th>pt13</th>\n",
       "      <th>eta13</th>\n",
       "      <th>phi13</th>\n",
       "      <th>E14</th>\n",
       "      <th>pt14</th>\n",
       "      <th>eta14</th>\n",
       "      <th>phi14</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>165476</td>\n",
       "      <td>165097</td>\n",
       "      <td>-0.0171131</td>\n",
       "      <td>-1.54448</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>\n",
       "      <td>...</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>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>105746</td>\n",
       "      <td>95333.9</td>\n",
       "      <td>0.449145</td>\n",
       "      <td>1.79273</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>\n",
       "      <td>...</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>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>436613</td>\n",
       "      <td>432317</td>\n",
       "      <td>-0.110325</td>\n",
       "      <td>0.656242</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>\n",
       "      <td>...</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>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>265871</td>\n",
       "      <td>210909</td>\n",
       "      <td>-0.671705</td>\n",
       "      <td>-2.92575</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>\n",
       "      <td>...</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>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>296867</td>\n",
       "      <td>102488</td>\n",
       "      <td>-1.72479</td>\n",
       "      <td>1.81765</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>\n",
       "      <td>...</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>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3276</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>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.559472</td>\n",
       "      <td>1.26894</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>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5436</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>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0393824</td>\n",
       "      <td>-1.53926</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>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5713</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>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.0629869</td>\n",
       "      <td>2.72413</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>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>263</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>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>42197.8</td>\n",
       "      <td>25092.7</td>\n",
       "      <td>-1.10318</td>\n",
       "      <td>0.640652</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1619</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>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>36741.3</td>\n",
       "      <td>35040.1</td>\n",
       "      <td>0.23484</td>\n",
       "      <td>1.77787</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>25555 rows × 56 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          E1      pt1        eta1      phi1   E2  pt2 eta2 phi2   E3  pt3  \\\n",
       "0     165476   165097  -0.0171131  -1.54448  NaN  NaN  NaN  NaN  NaN  NaN   \n",
       "1     105746  95333.9    0.449145   1.79273  NaN  NaN  NaN  NaN  NaN  NaN   \n",
       "4     436613   432317   -0.110325  0.656242  NaN  NaN  NaN  NaN  NaN  NaN   \n",
       "5     265871   210909   -0.671705  -2.92575  NaN  NaN  NaN  NaN  NaN  NaN   \n",
       "6     296867   102488    -1.72479   1.81765  NaN  NaN  NaN  NaN  NaN  NaN   \n",
       "...      ...      ...         ...       ...  ...  ...  ...  ...  ...  ...   \n",
       "3276     NaN      NaN         NaN       NaN  NaN  NaN  NaN  NaN  NaN  NaN   \n",
       "5436     NaN      NaN         NaN       NaN  NaN  NaN  NaN  NaN  NaN  NaN   \n",
       "5713     NaN      NaN         NaN       NaN  NaN  NaN  NaN  NaN  NaN  NaN   \n",
       "263      NaN      NaN         NaN       NaN  NaN  NaN  NaN  NaN  NaN  NaN   \n",
       "1619     NaN      NaN         NaN       NaN  NaN  NaN  NaN  NaN  NaN  NaN   \n",
       "\n",
       "      ...       eta12     phi12      E13     pt13     eta13     phi13 E14  \\\n",
       "0     ...         NaN       NaN      NaN      NaN       NaN       NaN NaN   \n",
       "1     ...         NaN       NaN      NaN      NaN       NaN       NaN NaN   \n",
       "4     ...         NaN       NaN      NaN      NaN       NaN       NaN NaN   \n",
       "5     ...         NaN       NaN      NaN      NaN       NaN       NaN NaN   \n",
       "6     ...         NaN       NaN      NaN      NaN       NaN       NaN NaN   \n",
       "...   ...         ...       ...      ...      ...       ...       ...  ..   \n",
       "3276  ...   -0.559472   1.26894      NaN      NaN       NaN       NaN NaN   \n",
       "5436  ...   0.0393824  -1.53926      NaN      NaN       NaN       NaN NaN   \n",
       "5713  ...  -0.0629869   2.72413      NaN      NaN       NaN       NaN NaN   \n",
       "263   ...         NaN       NaN  42197.8  25092.7  -1.10318  0.640652 NaN   \n",
       "1619  ...         NaN       NaN  36741.3  35040.1   0.23484   1.77787 NaN   \n",
       "\n",
       "     pt14 eta14 phi14  \n",
       "0     NaN   NaN   NaN  \n",
       "1     NaN   NaN   NaN  \n",
       "4     NaN   NaN   NaN  \n",
       "5     NaN   NaN   NaN  \n",
       "6     NaN   NaN   NaN  \n",
       "...   ...   ...   ...  \n",
       "3276  NaN   NaN   NaN  \n",
       "5436  NaN   NaN   NaN  \n",
       "5713  NaN   NaN   NaN  \n",
       "263   NaN   NaN   NaN  \n",
       "1619  NaN   NaN   NaN  \n",
       "\n",
       "[25555 rows x 56 columns]"
      ]
     },
     "execution_count": 88,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = data.fillna(-999)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [],
   "source": [
    "data['process_ID'] = 'ttbarZ'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>E1</th>\n",
       "      <th>pt1</th>\n",
       "      <th>eta1</th>\n",
       "      <th>phi1</th>\n",
       "      <th>E2</th>\n",
       "      <th>pt2</th>\n",
       "      <th>eta2</th>\n",
       "      <th>phi2</th>\n",
       "      <th>E3</th>\n",
       "      <th>pt3</th>\n",
       "      <th>...</th>\n",
       "      <th>phi12</th>\n",
       "      <th>E13</th>\n",
       "      <th>pt13</th>\n",
       "      <th>eta13</th>\n",
       "      <th>phi13</th>\n",
       "      <th>E14</th>\n",
       "      <th>pt14</th>\n",
       "      <th>eta14</th>\n",
       "      <th>phi14</th>\n",
       "      <th>process_ID</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>165476</td>\n",
       "      <td>165097</td>\n",
       "      <td>-0.0171131</td>\n",
       "      <td>-1.54448</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>...</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>ttbarZ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>105746</td>\n",
       "      <td>95333.9</td>\n",
       "      <td>0.449145</td>\n",
       "      <td>1.79273</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>...</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>ttbarZ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>436613</td>\n",
       "      <td>432317</td>\n",
       "      <td>-0.110325</td>\n",
       "      <td>0.656242</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>...</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>ttbarZ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>265871</td>\n",
       "      <td>210909</td>\n",
       "      <td>-0.671705</td>\n",
       "      <td>-2.92575</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>...</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>ttbarZ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>296867</td>\n",
       "      <td>102488</td>\n",
       "      <td>-1.72479</td>\n",
       "      <td>1.81765</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>...</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>ttbarZ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3276</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>...</td>\n",
       "      <td>1.26894</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>ttbarZ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5436</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>...</td>\n",
       "      <td>-1.53926</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>ttbarZ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5713</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>...</td>\n",
       "      <td>2.72413</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>ttbarZ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>263</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>...</td>\n",
       "      <td>-999</td>\n",
       "      <td>42197.8</td>\n",
       "      <td>25092.7</td>\n",
       "      <td>-1.10318</td>\n",
       "      <td>0.640652</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>ttbarZ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1619</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>...</td>\n",
       "      <td>-999</td>\n",
       "      <td>36741.3</td>\n",
       "      <td>35040.1</td>\n",
       "      <td>0.23484</td>\n",
       "      <td>1.77787</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>ttbarZ</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>25555 rows × 57 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          E1      pt1        eta1      phi1    E2   pt2  eta2  phi2    E3  \\\n",
       "0     165476   165097  -0.0171131  -1.54448  -999  -999  -999  -999  -999   \n",
       "1     105746  95333.9    0.449145   1.79273  -999  -999  -999  -999  -999   \n",
       "4     436613   432317   -0.110325  0.656242  -999  -999  -999  -999  -999   \n",
       "5     265871   210909   -0.671705  -2.92575  -999  -999  -999  -999  -999   \n",
       "6     296867   102488    -1.72479   1.81765  -999  -999  -999  -999  -999   \n",
       "...      ...      ...         ...       ...   ...   ...   ...   ...   ...   \n",
       "3276    -999     -999        -999      -999  -999  -999  -999  -999  -999   \n",
       "5436    -999     -999        -999      -999  -999  -999  -999  -999  -999   \n",
       "5713    -999     -999        -999      -999  -999  -999  -999  -999  -999   \n",
       "263     -999     -999        -999      -999  -999  -999  -999  -999  -999   \n",
       "1619    -999     -999        -999      -999  -999  -999  -999  -999  -999   \n",
       "\n",
       "       pt3  ...     phi12      E13     pt13     eta13     phi13    E14   pt14  \\\n",
       "0     -999  ...      -999     -999     -999      -999      -999 -999.0 -999.0   \n",
       "1     -999  ...      -999     -999     -999      -999      -999 -999.0 -999.0   \n",
       "4     -999  ...      -999     -999     -999      -999      -999 -999.0 -999.0   \n",
       "5     -999  ...      -999     -999     -999      -999      -999 -999.0 -999.0   \n",
       "6     -999  ...      -999     -999     -999      -999      -999 -999.0 -999.0   \n",
       "...    ...  ...       ...      ...      ...       ...       ...    ...    ...   \n",
       "3276  -999  ...   1.26894     -999     -999      -999      -999 -999.0 -999.0   \n",
       "5436  -999  ...  -1.53926     -999     -999      -999      -999 -999.0 -999.0   \n",
       "5713  -999  ...   2.72413     -999     -999      -999      -999 -999.0 -999.0   \n",
       "263   -999  ...      -999  42197.8  25092.7  -1.10318  0.640652 -999.0 -999.0   \n",
       "1619  -999  ...      -999  36741.3  35040.1   0.23484   1.77787 -999.0 -999.0   \n",
       "\n",
       "      eta14  phi14 process_ID  \n",
       "0    -999.0 -999.0     ttbarZ  \n",
       "1    -999.0 -999.0     ttbarZ  \n",
       "4    -999.0 -999.0     ttbarZ  \n",
       "5    -999.0 -999.0     ttbarZ  \n",
       "6    -999.0 -999.0     ttbarZ  \n",
       "...     ...    ...        ...  \n",
       "3276 -999.0 -999.0     ttbarZ  \n",
       "5436 -999.0 -999.0     ttbarZ  \n",
       "5713 -999.0 -999.0     ttbarZ  \n",
       "263  -999.0 -999.0     ttbarZ  \n",
       "1619 -999.0 -999.0     ttbarZ  \n",
       "\n",
       "[25555 rows x 57 columns]"
      ]
     },
     "execution_count": 91,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {},
   "outputs": [],
   "source": [
    "one_hot = pd.get_dummies(data['process_ID'])\n",
    "processes = one_hot.columns\n",
    "data.drop('process_ID', axis = 'columns', inplace = True)\n",
    "data = pd.concat([data, one_hot], sort = False, axis = 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>E1</th>\n",
       "      <th>pt1</th>\n",
       "      <th>eta1</th>\n",
       "      <th>phi1</th>\n",
       "      <th>E2</th>\n",
       "      <th>pt2</th>\n",
       "      <th>eta2</th>\n",
       "      <th>phi2</th>\n",
       "      <th>E3</th>\n",
       "      <th>pt3</th>\n",
       "      <th>...</th>\n",
       "      <th>phi12</th>\n",
       "      <th>E13</th>\n",
       "      <th>pt13</th>\n",
       "      <th>eta13</th>\n",
       "      <th>phi13</th>\n",
       "      <th>E14</th>\n",
       "      <th>pt14</th>\n",
       "      <th>eta14</th>\n",
       "      <th>phi14</th>\n",
       "      <th>ttbarZ</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>165476</td>\n",
       "      <td>165097</td>\n",
       "      <td>-0.0171131</td>\n",
       "      <td>-1.54448</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>...</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>105746</td>\n",
       "      <td>95333.9</td>\n",
       "      <td>0.449145</td>\n",
       "      <td>1.79273</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>...</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>436613</td>\n",
       "      <td>432317</td>\n",
       "      <td>-0.110325</td>\n",
       "      <td>0.656242</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>...</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>265871</td>\n",
       "      <td>210909</td>\n",
       "      <td>-0.671705</td>\n",
       "      <td>-2.92575</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>...</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>296867</td>\n",
       "      <td>102488</td>\n",
       "      <td>-1.72479</td>\n",
       "      <td>1.81765</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>...</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3276</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>...</td>\n",
       "      <td>1.26894</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5436</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>...</td>\n",
       "      <td>-1.53926</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5713</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>...</td>\n",
       "      <td>2.72413</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>263</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>...</td>\n",
       "      <td>-999</td>\n",
       "      <td>42197.8</td>\n",
       "      <td>25092.7</td>\n",
       "      <td>-1.10318</td>\n",
       "      <td>0.640652</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1619</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>-999</td>\n",
       "      <td>...</td>\n",
       "      <td>-999</td>\n",
       "      <td>36741.3</td>\n",
       "      <td>35040.1</td>\n",
       "      <td>0.23484</td>\n",
       "      <td>1.77787</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>-999.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>25555 rows × 57 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          E1      pt1        eta1      phi1    E2   pt2  eta2  phi2    E3  \\\n",
       "0     165476   165097  -0.0171131  -1.54448  -999  -999  -999  -999  -999   \n",
       "1     105746  95333.9    0.449145   1.79273  -999  -999  -999  -999  -999   \n",
       "4     436613   432317   -0.110325  0.656242  -999  -999  -999  -999  -999   \n",
       "5     265871   210909   -0.671705  -2.92575  -999  -999  -999  -999  -999   \n",
       "6     296867   102488    -1.72479   1.81765  -999  -999  -999  -999  -999   \n",
       "...      ...      ...         ...       ...   ...   ...   ...   ...   ...   \n",
       "3276    -999     -999        -999      -999  -999  -999  -999  -999  -999   \n",
       "5436    -999     -999        -999      -999  -999  -999  -999  -999  -999   \n",
       "5713    -999     -999        -999      -999  -999  -999  -999  -999  -999   \n",
       "263     -999     -999        -999      -999  -999  -999  -999  -999  -999   \n",
       "1619    -999     -999        -999      -999  -999  -999  -999  -999  -999   \n",
       "\n",
       "       pt3  ...     phi12      E13     pt13     eta13     phi13    E14   pt14  \\\n",
       "0     -999  ...      -999     -999     -999      -999      -999 -999.0 -999.0   \n",
       "1     -999  ...      -999     -999     -999      -999      -999 -999.0 -999.0   \n",
       "4     -999  ...      -999     -999     -999      -999      -999 -999.0 -999.0   \n",
       "5     -999  ...      -999     -999     -999      -999      -999 -999.0 -999.0   \n",
       "6     -999  ...      -999     -999     -999      -999      -999 -999.0 -999.0   \n",
       "...    ...  ...       ...      ...      ...       ...       ...    ...    ...   \n",
       "3276  -999  ...   1.26894     -999     -999      -999      -999 -999.0 -999.0   \n",
       "5436  -999  ...  -1.53926     -999     -999      -999      -999 -999.0 -999.0   \n",
       "5713  -999  ...   2.72413     -999     -999      -999      -999 -999.0 -999.0   \n",
       "263   -999  ...      -999  42197.8  25092.7  -1.10318  0.640652 -999.0 -999.0   \n",
       "1619  -999  ...      -999  36741.3  35040.1   0.23484   1.77787 -999.0 -999.0   \n",
       "\n",
       "      eta14  phi14 ttbarZ  \n",
       "0    -999.0 -999.0      1  \n",
       "1    -999.0 -999.0      1  \n",
       "4    -999.0 -999.0      1  \n",
       "5    -999.0 -999.0      1  \n",
       "6    -999.0 -999.0      1  \n",
       "...     ...    ...    ...  \n",
       "3276 -999.0 -999.0      1  \n",
       "5436 -999.0 -999.0      1  \n",
       "5713 -999.0 -999.0      1  \n",
       "263  -999.0 -999.0      1  \n",
       "1619 -999.0 -999.0      1  \n",
       "\n",
       "[25555 rows x 57 columns]"
      ]
     },
     "execution_count": 93,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [],
   "source": [
    "variables = [entry for entry in df.columns if entry[0] == 'E'] + [entry for entry in df.columns if entry[0:2] == 'pt'] + [entry for entry in df.columns if entry[0:2] == 'et'] + [entry for entry in df.columns if entry[0:2] == 'ph']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [],
   "source": [
    "#We have created a df of our values and some kind of class label\n",
    "#We should normalize\n",
    "x = data[variables].values\n",
    "x_scaled = StandardScaler().fit_transform(x)\n",
    "data[variables] = x_scaled\n",
    "\n",
    "x_train, x_test, y_train, y_test = train_test_split(data[variables].values,\n",
    "\t\t\t\t\t\t    data[processes].values, \n",
    "                                                    shuffle = True,\n",
    "                                                    random_state = 42,\n",
    "                                                    test_size = 0.1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Reshape the data. This will have to be done explicitly until you think of a clever way of doing it\n",
    "x_train = x_train.reshape((x_train.shape[0], 4, 2, 7))\n",
    "x_test = x_test.reshape((x_test.shape[0], 4, 2, 7))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [],
   "source": [
    "def sampling(args):\n",
    "    z_mean, z_log_var=args\n",
    "    epsilon=tf.random.uniform(shape=(K.shape(z_mean)[0], latent_dim))\n",
    "    return z_mean+K.exp(z_log_var/2)*epsilon"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {},
   "outputs": [],
   "source": [
    "#VAE Model\n",
    "original_dim = x_train.shape[0]\n",
    "input_shape = x_train.shape[1:]\n",
    "latent_dim = 4\n",
    "intermediate_dim = 50\n",
    "kernel_max_norm = 1000.\n",
    "act_fun = 'relu'\n",
    "epsilon_std = 1.\n",
    "filters = 32\n",
    "kernel_size = (3, 3)\n",
    "strides = (1, 1)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Layers\n",
    "x_input = Input(shape = input_shape, name = 'Input')\n",
    "conv_1 = Conv2D(filters = filters, \n",
    "                activation = act_fun,\n",
    "                kernel_size = kernel_size,\n",
    "                strides = strides, \n",
    "                padding = 'same')(x_input)\n",
    "pool_1 = MaxPooling2D(pool_size = (1, 1),\n",
    "                      strides = 2,\n",
    "                      #data_format = 'channels_first'\n",
    "                     )(conv_1)\n",
    "conv_2 = Conv2D(filters = filters, \n",
    "                activation = act_fun, \n",
    "                kernel_size = kernel_size, \n",
    "                strides = strides, \n",
    "                padding = 'same')(pool_1)\n",
    "\n",
    "#Shape info needed to inform the dense layer in the decoder\n",
    "flatten = Flatten()(conv_2)\n",
    "#cnn = Model(x_input, flatten, name = 'cnn')\n",
    "#cnn.summary()\n",
    "\n",
    "shape = input_shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"encoder\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "Input (InputLayer)              (None, 4, 2, 7)      0                                            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_5 (Conv2D)               (None, 4, 2, 32)     2048        Input[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling2d_3 (MaxPooling2D)  (None, 2, 1, 32)     0           conv2d_5[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_6 (Conv2D)               (None, 2, 1, 32)     9248        max_pooling2d_3[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "flatten_3 (Flatten)             (None, 64)           0           conv2d_6[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "dense_encoder_1 (Dense)         (None, 50)           3250        flatten_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dense_encoder_2 (Dense)         (None, 16)           816         dense_encoder_1[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "z_mean (Dense)                  (None, 4)            68          dense_encoder_2[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "z_var (Dense)                   (None, 4)            68          dense_encoder_2[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "sampling (Lambda)               (None, 4)            0           z_mean[0][0]                     \n",
      "                                                                 z_var[0][0]                      \n",
      "==================================================================================================\n",
      "Total params: 15,498\n",
      "Trainable params: 15,498\n",
      "Non-trainable params: 0\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "#Encoding\n",
    "dense_encoder_1 = Dense(intermediate_dim,\n",
    "                        activation = act_fun,\n",
    "                        name = 'dense_encoder_1')(flatten)\n",
    "dense_encoder_2 = Dense(16,\n",
    "                        activation = act_fun,\n",
    "                        name = 'dense_encoder_2')(dense_encoder_1)\n",
    "z_mean = Dense(latent_dim, name = 'z_mean')(dense_encoder_2)\n",
    "z_var = Dense(latent_dim, name = 'z_var')(dense_encoder_2)\n",
    "\n",
    "z = Lambda(sampling, output_shape = (latent_dim, ), name = 'sampling')([z_mean, z_var])\n",
    "\n",
    "encoder = Model(x_input, [z_mean, z_var, z], name = 'encoder')\n",
    "encoder.summary()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"decoder\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "decoder_input (InputLayer)   (None, 4)                 0         \n",
      "_________________________________________________________________\n",
      "dense_decoder_1 (Dense)      (None, 56)                280       \n",
      "_________________________________________________________________\n",
      "reshape_3 (Reshape)          (None, 4, 2, 7)           0         \n",
      "_________________________________________________________________\n",
      "deconv_1 (Conv2DTranspose)   (None, 4, 2, 32)          2048      \n",
      "_________________________________________________________________\n",
      "deconv_2 (Conv2DTranspose)   (None, 4, 2, 16)          4624      \n",
      "_________________________________________________________________\n",
      "decoder_output (Conv2DTransp (None, 4, 2, 7)           1015      \n",
      "=================================================================\n",
      "Total params: 7,967\n",
      "Trainable params: 7,967\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "#Decoding\n",
    "decoder_input = Input(shape = (latent_dim, ), name = 'decoder_input')\n",
    "dense_decoder_1 = Dense(shape[0]*shape[1]*shape[2], \n",
    "                        activation = act_fun, name = 'dense_decoder_1')(decoder_input)\n",
    "reshape = Reshape((shape[0], shape[1], shape[2]))(dense_decoder_1)\n",
    "\n",
    "deconv_1 = Conv2DTranspose(filters = filters,\n",
    "                           kernel_size = kernel_size, \n",
    "                           activation = act_fun, \n",
    "                           strides = strides, \n",
    "                           padding = 'same', \n",
    "                           name = 'deconv_1')(reshape)\n",
    "deconv_2 = Conv2DTranspose(filters = 16,\n",
    "                           kernel_size = kernel_size, \n",
    "                           activation = act_fun, \n",
    "                           strides = strides, \n",
    "                           padding = 'same', \n",
    "                           name = 'deconv_2')(deconv_1)\n",
    "\n",
    "output = Conv2DTranspose(filters = 7, \n",
    "                         kernel_size = kernel_size, \n",
    "                         activation = 'sigmoid', \n",
    "                         padding = 'same', \n",
    "                         strides = strides, \n",
    "                         name = 'decoder_output')(deconv_2)\n",
    "\n",
    "decoder = Model(decoder_input, output, name = 'decoder')\n",
    "decoder.summary()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Build the VAE model\n",
    "outputs = decoder(encoder(x_input)[2])\n",
    "vae = Model(x_input, outputs, name = 'vae')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Define some losses\n",
    "def kl_loss(y_true, y_pred):\n",
    "    kl_loss = 1 + z_var - K.square(z_mean) - K.exp(z_var)\n",
    "    kl_loss = K.sum(kl_loss, axis=-1)\n",
    "    kl_loss *= -0.5\n",
    "    #kl_loss = K.mean(kl_loss)\n",
    "    return kl_loss\n",
    "\n",
    "def recon_loss(y_true, y_pred):\n",
    "    recon_loss = mean_squared_error(K.flatten(y_true), K.flatten(y_pred))\n",
    "    recon_loss *= shape[0]*shape[1]*shape[2]\n",
    "    #recon_loss = K.mean(recon_loss)\n",
    "    return recon_loss\n",
    "\n",
    "def my_vae_loss(y_true, y_pred):\n",
    "    recon = recon_loss(y_true, y_pred) \n",
    "    kl = kl_loss(y_true, y_pred)\n",
    "    vae_loss = recon + kl\n",
    "    return vae_loss\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"vae\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "Input (InputLayer)           (None, 4, 2, 7)           0         \n",
      "_________________________________________________________________\n",
      "encoder (Model)              [(None, 4), (None, 4), (N 15498     \n",
      "_________________________________________________________________\n",
      "decoder (Model)              (None, 4, 2, 7)           7967      \n",
      "=================================================================\n",
      "Total params: 23,465\n",
      "Trainable params: 23,465\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "vae.compile(optimizer = 'rmsprop', loss = my_vae_loss, metrics = [kl_loss, recon_loss])\n",
    "vae.summary()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 22999 samples, validate on 22999 samples\n",
      "Epoch 1/100\n",
      " - 1s - loss: 51.9614 - kl_loss: 0.2475 - recon_loss: 51.7134 - val_loss: 49.1760 - val_kl_loss: 0.3952 - val_recon_loss: 48.7800\n",
      "Epoch 2/100\n",
      " - 1s - loss: 48.3187 - kl_loss: 0.3636 - recon_loss: 47.9545 - val_loss: 47.7833 - val_kl_loss: 0.3083 - val_recon_loss: 47.4741\n",
      "Epoch 3/100\n",
      " - 1s - loss: 47.7086 - kl_loss: 0.2970 - recon_loss: 47.4112 - val_loss: 47.5253 - val_kl_loss: 0.2633 - val_recon_loss: 47.2612\n",
      "Epoch 4/100\n",
      " - 1s - loss: 47.5771 - kl_loss: 0.2568 - recon_loss: 47.3198 - val_loss: 47.4630 - val_kl_loss: 0.2507 - val_recon_loss: 47.2115\n",
      "Epoch 5/100\n",
      " - 1s - loss: 47.4678 - kl_loss: 0.2407 - recon_loss: 47.2273 - val_loss: 47.3516 - val_kl_loss: 0.2147 - val_recon_loss: 47.1361\n",
      "Epoch 6/100\n",
      " - 1s - loss: 47.2924 - kl_loss: 0.2464 - recon_loss: 47.0556 - val_loss: 47.1616 - val_kl_loss: 0.2274 - val_recon_loss: 46.9334\n",
      "Epoch 7/100\n",
      " - 1s - loss: 47.2089 - kl_loss: 0.2365 - recon_loss: 46.9718 - val_loss: 47.1176 - val_kl_loss: 0.2092 - val_recon_loss: 46.9076\n",
      "Epoch 8/100\n",
      " - 1s - loss: 46.8855 - kl_loss: 0.2508 - recon_loss: 46.6342 - val_loss: 46.7015 - val_kl_loss: 0.2465 - val_recon_loss: 46.4542\n",
      "Epoch 9/100\n",
      " - 1s - loss: 46.7125 - kl_loss: 0.2261 - recon_loss: 46.4864 - val_loss: 46.8862 - val_kl_loss: 0.2172 - val_recon_loss: 46.6681\n",
      "Epoch 10/100\n",
      " - 1s - loss: 46.6663 - kl_loss: 0.2216 - recon_loss: 46.4441 - val_loss: 46.6451 - val_kl_loss: 0.2275 - val_recon_loss: 46.4168\n",
      "Epoch 11/100\n",
      " - 1s - loss: 46.6520 - kl_loss: 0.2177 - recon_loss: 46.4338 - val_loss: 46.6383 - val_kl_loss: 0.2295 - val_recon_loss: 46.4080\n",
      "Epoch 12/100\n",
      " - 1s - loss: 46.6267 - kl_loss: 0.2051 - recon_loss: 46.4208 - val_loss: 46.5963 - val_kl_loss: 0.1987 - val_recon_loss: 46.3968\n",
      "Epoch 13/100\n",
      " - 1s - loss: 46.6144 - kl_loss: 0.2055 - recon_loss: 46.4082 - val_loss: 46.5898 - val_kl_loss: 0.2003 - val_recon_loss: 46.3887\n",
      "Epoch 14/100\n",
      " - 1s - loss: 46.6321 - kl_loss: 0.2041 - recon_loss: 46.4278 - val_loss: 46.5690 - val_kl_loss: 0.1910 - val_recon_loss: 46.3772\n",
      "Epoch 15/100\n",
      " - 1s - loss: 46.6286 - kl_loss: 0.1945 - recon_loss: 46.4387 - val_loss: 46.5849 - val_kl_loss: 0.1995 - val_recon_loss: 46.3846\n",
      "Epoch 16/100\n",
      " - 1s - loss: 46.6181 - kl_loss: 0.1935 - recon_loss: 46.4241 - val_loss: 46.6156 - val_kl_loss: 0.2323 - val_recon_loss: 46.3825\n",
      "Epoch 17/100\n",
      " - 1s - loss: 46.6092 - kl_loss: 0.1976 - recon_loss: 46.4111 - val_loss: 46.5659 - val_kl_loss: 0.1873 - val_recon_loss: 46.3778\n",
      "Epoch 18/100\n",
      " - 1s - loss: 46.6206 - kl_loss: 0.1924 - recon_loss: 46.4324 - val_loss: 46.5804 - val_kl_loss: 0.1957 - val_recon_loss: 46.3839\n",
      "Epoch 19/100\n",
      " - 1s - loss: 46.6161 - kl_loss: 0.1799 - recon_loss: 46.4353 - val_loss: 46.5885 - val_kl_loss: 0.1931 - val_recon_loss: 46.3946\n",
      "Epoch 20/100\n",
      " - 1s - loss: 46.6278 - kl_loss: 0.1884 - recon_loss: 46.4389 - val_loss: 46.5631 - val_kl_loss: 0.1922 - val_recon_loss: 46.3701\n",
      "Epoch 21/100\n",
      " - 1s - loss: 46.6031 - kl_loss: 0.1854 - recon_loss: 46.4169 - val_loss: 46.5696 - val_kl_loss: 0.1941 - val_recon_loss: 46.3746\n",
      "Epoch 22/100\n",
      " - 1s - loss: 46.5952 - kl_loss: 0.1839 - recon_loss: 46.4105 - val_loss: 46.7865 - val_kl_loss: 0.1717 - val_recon_loss: 46.6140\n",
      "Epoch 23/100\n",
      " - 1s - loss: 46.5975 - kl_loss: 0.1787 - recon_loss: 46.4186 - val_loss: 46.5400 - val_kl_loss: 0.1762 - val_recon_loss: 46.3630\n",
      "Epoch 24/100\n",
      " - 1s - loss: 46.5962 - kl_loss: 0.1777 - recon_loss: 46.4180 - val_loss: 46.5320 - val_kl_loss: 0.1508 - val_recon_loss: 46.3805\n",
      "Epoch 25/100\n",
      " - 1s - loss: 46.6027 - kl_loss: 0.1733 - recon_loss: 46.4288 - val_loss: 46.5488 - val_kl_loss: 0.1752 - val_recon_loss: 46.3727\n",
      "Epoch 26/100\n",
      " - 1s - loss: 46.6268 - kl_loss: 0.1784 - recon_loss: 46.4480 - val_loss: 46.5906 - val_kl_loss: 0.1721 - val_recon_loss: 46.4178\n",
      "Epoch 27/100\n",
      " - 1s - loss: 46.6208 - kl_loss: 0.1733 - recon_loss: 46.4469 - val_loss: 46.5806 - val_kl_loss: 0.1861 - val_recon_loss: 46.3937\n",
      "Epoch 28/100\n",
      " - 1s - loss: 46.5910 - kl_loss: 0.1726 - recon_loss: 46.4177 - val_loss: 46.5531 - val_kl_loss: 0.1636 - val_recon_loss: 46.3886\n",
      "Epoch 29/100\n",
      " - 1s - loss: 46.6217 - kl_loss: 0.1674 - recon_loss: 46.4540 - val_loss: 46.6555 - val_kl_loss: 0.1726 - val_recon_loss: 46.4821\n",
      "Epoch 30/100\n",
      " - 1s - loss: 46.6097 - kl_loss: 0.1644 - recon_loss: 46.4447 - val_loss: 46.5948 - val_kl_loss: 0.1591 - val_recon_loss: 46.4349\n",
      "Epoch 31/100\n",
      " - 1s - loss: 46.5949 - kl_loss: 0.1701 - recon_loss: 46.4242 - val_loss: 46.6366 - val_kl_loss: 0.1683 - val_recon_loss: 46.4675\n",
      "Epoch 32/100\n",
      " - 1s - loss: 46.5909 - kl_loss: 0.1653 - recon_loss: 46.4255 - val_loss: 46.6312 - val_kl_loss: 0.1942 - val_recon_loss: 46.4362\n",
      "Epoch 33/100\n",
      " - 1s - loss: 46.6252 - kl_loss: 0.1724 - recon_loss: 46.4529 - val_loss: 46.5810 - val_kl_loss: 0.1581 - val_recon_loss: 46.4220\n",
      "Epoch 34/100\n",
      " - 1s - loss: 46.6354 - kl_loss: 0.1755 - recon_loss: 46.4598 - val_loss: 46.5763 - val_kl_loss: 0.1800 - val_recon_loss: 46.3954\n",
      "Epoch 35/100\n",
      " - 1s - loss: 46.6090 - kl_loss: 0.1723 - recon_loss: 46.4368 - val_loss: 46.5708 - val_kl_loss: 0.1727 - val_recon_loss: 46.3973\n",
      "Epoch 36/100\n",
      " - 1s - loss: 46.6233 - kl_loss: 0.1931 - recon_loss: 46.4299 - val_loss: 46.5303 - val_kl_loss: 0.1510 - val_recon_loss: 46.3785\n",
      "Epoch 37/100\n",
      " - 1s - loss: 46.5985 - kl_loss: 0.1785 - recon_loss: 46.4206 - val_loss: 46.5863 - val_kl_loss: 0.1762 - val_recon_loss: 46.4093\n",
      "Epoch 38/100\n",
      " - 1s - loss: 46.6512 - kl_loss: 0.1864 - recon_loss: 46.4649 - val_loss: 46.5979 - val_kl_loss: 0.1563 - val_recon_loss: 46.4408\n",
      "Epoch 39/100\n",
      " - 1s - loss: 46.6638 - kl_loss: 0.1847 - recon_loss: 46.4783 - val_loss: 46.7895 - val_kl_loss: 0.1636 - val_recon_loss: 46.6250\n",
      "Epoch 40/100\n",
      " - 1s - loss: 46.7038 - kl_loss: 0.1804 - recon_loss: 46.5226 - val_loss: 46.6433 - val_kl_loss: 0.1888 - val_recon_loss: 46.4537\n",
      "Epoch 41/100\n",
      " - 1s - loss: 46.7119 - kl_loss: 0.2961 - recon_loss: 46.4158 - val_loss: 46.8817 - val_kl_loss: 0.1806 - val_recon_loss: 46.7002\n",
      "Epoch 42/100\n",
      " - 1s - loss: 46.6577 - kl_loss: 0.1877 - recon_loss: 46.4706 - val_loss: 46.6390 - val_kl_loss: 0.2182 - val_recon_loss: 46.4200\n",
      "Epoch 43/100\n",
      " - 1s - loss: 46.6283 - kl_loss: 0.1911 - recon_loss: 46.4369 - val_loss: 46.5712 - val_kl_loss: 0.2027 - val_recon_loss: 46.3677\n",
      "Epoch 44/100\n",
      " - 1s - loss: 46.6457 - kl_loss: 0.1882 - recon_loss: 46.4567 - val_loss: 46.6194 - val_kl_loss: 0.1735 - val_recon_loss: 46.4451\n",
      "Epoch 45/100\n",
      " - 1s - loss: 46.7326 - kl_loss: 0.1923 - recon_loss: 46.5405 - val_loss: 46.6874 - val_kl_loss: 0.1670 - val_recon_loss: 46.5195\n",
      "Epoch 46/100\n",
      " - 1s - loss: 46.6844 - kl_loss: 0.1850 - recon_loss: 46.4986 - val_loss: 46.5734 - val_kl_loss: 0.1645 - val_recon_loss: 46.4081\n",
      "Epoch 47/100\n",
      " - 1s - loss: 46.6456 - kl_loss: 0.1849 - recon_loss: 46.4611 - val_loss: 46.6200 - val_kl_loss: 0.1965 - val_recon_loss: 46.4226\n",
      "Epoch 48/100\n",
      " - 1s - loss: 46.6410 - kl_loss: 0.1937 - recon_loss: 46.4468 - val_loss: 46.7699 - val_kl_loss: 0.2106 - val_recon_loss: 46.5585\n",
      "Epoch 49/100\n",
      " - 1s - loss: 46.6627 - kl_loss: 0.1915 - recon_loss: 46.4705 - val_loss: 46.5833 - val_kl_loss: 0.1823 - val_recon_loss: 46.4002\n",
      "Epoch 50/100\n",
      " - 1s - loss: 46.7050 - kl_loss: 0.1970 - recon_loss: 46.5074 - val_loss: 46.7150 - val_kl_loss: 0.1794 - val_recon_loss: 46.5348\n",
      "Epoch 51/100\n",
      " - 1s - loss: 46.6718 - kl_loss: 0.1986 - recon_loss: 46.4724 - val_loss: 46.7176 - val_kl_loss: 0.1788 - val_recon_loss: 46.5380\n",
      "Epoch 52/100\n",
      " - 1s - loss: 46.6459 - kl_loss: 0.1967 - recon_loss: 46.4496 - val_loss: 46.5901 - val_kl_loss: 0.2007 - val_recon_loss: 46.3886\n",
      "Epoch 53/100\n",
      " - 1s - loss: 46.7942 - kl_loss: 0.1986 - recon_loss: 46.5949 - val_loss: 47.8303 - val_kl_loss: 0.2310 - val_recon_loss: 47.5985\n",
      "Epoch 54/100\n",
      " - 1s - loss: 46.9693 - kl_loss: 0.2027 - recon_loss: 46.7656 - val_loss: 46.9437 - val_kl_loss: 0.1591 - val_recon_loss: 46.7837\n",
      "Epoch 55/100\n",
      " - 1s - loss: 46.9928 - kl_loss: 0.2231 - recon_loss: 46.7717 - val_loss: 46.8761 - val_kl_loss: 0.1922 - val_recon_loss: 46.6830\n",
      "Epoch 56/100\n",
      " - 1s - loss: 46.9489 - kl_loss: 0.2210 - recon_loss: 46.7275 - val_loss: 47.0747 - val_kl_loss: 0.2428 - val_recon_loss: 46.8311\n",
      "Epoch 57/100\n",
      " - 1s - loss: 46.9198 - kl_loss: 0.2137 - recon_loss: 46.7069 - val_loss: 46.8816 - val_kl_loss: 0.2074 - val_recon_loss: 46.6733\n",
      "Epoch 58/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " - 1s - loss: 47.0004 - kl_loss: 0.2214 - recon_loss: 46.7780 - val_loss: 46.8699 - val_kl_loss: 0.2455 - val_recon_loss: 46.6235\n",
      "Epoch 59/100\n",
      " - 1s - loss: 46.9868 - kl_loss: 0.2296 - recon_loss: 46.7564 - val_loss: 46.9705 - val_kl_loss: 0.2122 - val_recon_loss: 46.7575\n",
      "Epoch 60/100\n",
      " - 1s - loss: 46.9689 - kl_loss: 0.2136 - recon_loss: 46.7550 - val_loss: 46.8801 - val_kl_loss: 0.2338 - val_recon_loss: 46.6454\n",
      "Epoch 61/100\n",
      " - 1s - loss: 46.9678 - kl_loss: 0.2112 - recon_loss: 46.7560 - val_loss: 47.2265 - val_kl_loss: 0.2471 - val_recon_loss: 46.9786\n",
      "Epoch 62/100\n",
      " - 1s - loss: 47.0218 - kl_loss: 0.2156 - recon_loss: 46.8071 - val_loss: 46.9715 - val_kl_loss: 0.2585 - val_recon_loss: 46.7122\n",
      "Epoch 63/100\n",
      " - 1s - loss: 46.9701 - kl_loss: 0.2423 - recon_loss: 46.7282 - val_loss: 47.1122 - val_kl_loss: 0.1893 - val_recon_loss: 46.9220\n",
      "Epoch 64/100\n",
      " - 1s - loss: 46.9388 - kl_loss: 0.2282 - recon_loss: 46.7097 - val_loss: 46.8549 - val_kl_loss: 0.2385 - val_recon_loss: 46.6155\n",
      "Epoch 65/100\n",
      " - 1s - loss: 46.9622 - kl_loss: 0.2248 - recon_loss: 46.7380 - val_loss: 46.8888 - val_kl_loss: 0.1960 - val_recon_loss: 46.6919\n",
      "Epoch 66/100\n",
      " - 1s - loss: 47.2822 - kl_loss: 0.5310 - recon_loss: 46.7506 - val_loss: 46.8737 - val_kl_loss: 0.2348 - val_recon_loss: 46.6380\n",
      "Epoch 67/100\n",
      " - 1s - loss: 46.9880 - kl_loss: 0.2198 - recon_loss: 46.7681 - val_loss: 46.8805 - val_kl_loss: 0.1774 - val_recon_loss: 46.7023\n",
      "Epoch 68/100\n",
      " - 1s - loss: 46.9494 - kl_loss: 0.2295 - recon_loss: 46.7196 - val_loss: 46.8415 - val_kl_loss: 0.2098 - val_recon_loss: 46.6309\n",
      "Epoch 69/100\n",
      " - 1s - loss: 46.9862 - kl_loss: 0.2135 - recon_loss: 46.7721 - val_loss: 47.0748 - val_kl_loss: 0.2115 - val_recon_loss: 46.8624\n",
      "Epoch 70/100\n",
      " - 1s - loss: 47.3449 - kl_loss: 0.2166 - recon_loss: 47.1283 - val_loss: 47.3793 - val_kl_loss: 0.2074 - val_recon_loss: 47.1711\n",
      "Epoch 71/100\n",
      " - 1s - loss: 47.4312 - kl_loss: 0.2115 - recon_loss: 47.2190 - val_loss: 47.3910 - val_kl_loss: 0.2456 - val_recon_loss: 47.1446\n",
      "Epoch 72/100\n",
      " - 1s - loss: 47.4017 - kl_loss: 0.2148 - recon_loss: 47.1863 - val_loss: 47.3134 - val_kl_loss: 0.2227 - val_recon_loss: 47.0898\n",
      "Epoch 73/100\n",
      " - 1s - loss: 47.4227 - kl_loss: 0.2106 - recon_loss: 47.2112 - val_loss: 47.3104 - val_kl_loss: 0.2278 - val_recon_loss: 47.0818\n",
      "Epoch 74/100\n",
      " - 1s - loss: 47.3835 - kl_loss: 0.2093 - recon_loss: 47.1746 - val_loss: 47.4786 - val_kl_loss: 0.1715 - val_recon_loss: 47.3063\n",
      "Epoch 75/100\n",
      " - 1s - loss: 47.4094 - kl_loss: 0.2004 - recon_loss: 47.2091 - val_loss: 47.3266 - val_kl_loss: 0.2527 - val_recon_loss: 47.0730\n",
      "Epoch 76/100\n",
      " - 1s - loss: 47.3923 - kl_loss: 0.2188 - recon_loss: 47.1727 - val_loss: 47.3815 - val_kl_loss: 0.2414 - val_recon_loss: 47.1392\n",
      "Epoch 77/100\n",
      " - 1s - loss: 47.3994 - kl_loss: 0.2050 - recon_loss: 47.1940 - val_loss: 47.3003 - val_kl_loss: 0.2154 - val_recon_loss: 47.0840\n",
      "Epoch 78/100\n",
      " - 1s - loss: 47.3927 - kl_loss: 0.2053 - recon_loss: 47.1868 - val_loss: 47.4380 - val_kl_loss: 0.1813 - val_recon_loss: 47.2559\n",
      "Epoch 79/100\n",
      " - 1s - loss: 47.4377 - kl_loss: 0.2079 - recon_loss: 47.2295 - val_loss: 47.3886 - val_kl_loss: 0.1627 - val_recon_loss: 47.2251\n",
      "Epoch 80/100\n",
      " - 1s - loss: 47.3920 - kl_loss: 0.2072 - recon_loss: 47.1840 - val_loss: 47.3498 - val_kl_loss: 0.2028 - val_recon_loss: 47.1462\n",
      "Epoch 81/100\n",
      " - 1s - loss: 47.4481 - kl_loss: 0.2190 - recon_loss: 47.2285 - val_loss: 47.3370 - val_kl_loss: 0.1804 - val_recon_loss: 47.1557\n",
      "Epoch 82/100\n",
      " - 1s - loss: 47.4352 - kl_loss: 0.2215 - recon_loss: 47.2137 - val_loss: 47.6854 - val_kl_loss: 0.1413 - val_recon_loss: 47.5432\n",
      "Epoch 83/100\n",
      " - 1s - loss: 47.4234 - kl_loss: 0.2143 - recon_loss: 47.2085 - val_loss: 47.3537 - val_kl_loss: 0.2008 - val_recon_loss: 47.1520\n",
      "Epoch 84/100\n",
      " - 1s - loss: 48.1007 - kl_loss: 0.8814 - recon_loss: 47.2209 - val_loss: 47.3384 - val_kl_loss: 0.1825 - val_recon_loss: 47.1550\n",
      "Epoch 85/100\n",
      " - 1s - loss: 47.4359 - kl_loss: 0.2250 - recon_loss: 47.2106 - val_loss: 47.5165 - val_kl_loss: 0.3084 - val_recon_loss: 47.2072\n",
      "Epoch 86/100\n",
      " - 1s - loss: 47.5226 - kl_loss: 0.3246 - recon_loss: 47.1980 - val_loss: 47.4127 - val_kl_loss: 0.2518 - val_recon_loss: 47.1601\n",
      "Epoch 87/100\n",
      " - 1s - loss: 47.4411 - kl_loss: 0.2297 - recon_loss: 47.2111 - val_loss: 47.3628 - val_kl_loss: 0.2212 - val_recon_loss: 47.1407\n",
      "Epoch 88/100\n",
      " - 1s - loss: 47.4860 - kl_loss: 0.2368 - recon_loss: 47.2487 - val_loss: 47.3800 - val_kl_loss: 0.2459 - val_recon_loss: 47.1332\n",
      "Epoch 89/100\n",
      " - 1s - loss: 47.4124 - kl_loss: 0.2300 - recon_loss: 47.1821 - val_loss: 47.4430 - val_kl_loss: 0.2140 - val_recon_loss: 47.2281\n",
      "Epoch 90/100\n",
      " - 1s - loss: 47.4397 - kl_loss: 0.2205 - recon_loss: 47.2187 - val_loss: 47.3505 - val_kl_loss: 0.1572 - val_recon_loss: 47.1925\n",
      "Epoch 91/100\n",
      " - 1s - loss: 47.4176 - kl_loss: 0.2077 - recon_loss: 47.2095 - val_loss: 47.4140 - val_kl_loss: 0.1813 - val_recon_loss: 47.2319\n",
      "Epoch 92/100\n",
      " - 1s - loss: 47.4329 - kl_loss: 0.2166 - recon_loss: 47.2182 - val_loss: 47.3528 - val_kl_loss: 0.2409 - val_recon_loss: 47.1110\n",
      "Epoch 93/100\n",
      " - 1s - loss: 47.5919 - kl_loss: 0.3629 - recon_loss: 47.2285 - val_loss: 47.3167 - val_kl_loss: 0.1849 - val_recon_loss: 47.1309\n",
      "Epoch 94/100\n",
      " - 1s - loss: 47.4589 - kl_loss: 0.2049 - recon_loss: 47.2534 - val_loss: 47.3415 - val_kl_loss: 0.2091 - val_recon_loss: 47.1316\n",
      "Epoch 95/100\n",
      " - 1s - loss: 47.4345 - kl_loss: 0.2118 - recon_loss: 47.2222 - val_loss: 47.3126 - val_kl_loss: 0.2117 - val_recon_loss: 47.1000\n",
      "Epoch 96/100\n",
      " - 1s - loss: 47.4114 - kl_loss: 0.1999 - recon_loss: 47.2109 - val_loss: 47.4943 - val_kl_loss: 0.2018 - val_recon_loss: 47.2917\n",
      "Epoch 97/100\n",
      " - 1s - loss: 47.4028 - kl_loss: 0.1919 - recon_loss: 47.2113 - val_loss: 47.4460 - val_kl_loss: 0.1480 - val_recon_loss: 47.2971\n",
      "Epoch 98/100\n",
      " - 1s - loss: 47.4142 - kl_loss: 0.2156 - recon_loss: 47.1981 - val_loss: 47.3154 - val_kl_loss: 0.2252 - val_recon_loss: 47.0894\n",
      "Epoch 99/100\n",
      " - 1s - loss: 47.3847 - kl_loss: 0.1927 - recon_loss: 47.1927 - val_loss: 47.6386 - val_kl_loss: 0.1288 - val_recon_loss: 47.5089\n",
      "Epoch 100/100\n",
      " - 1s - loss: 47.4294 - kl_loss: 0.1961 - recon_loss: 47.2330 - val_loss: 47.3074 - val_kl_loss: 0.1927 - val_recon_loss: 47.1139\n"
     ]
    }
   ],
   "source": [
    "#Fit the model\n",
    "epochs = 100\n",
    "batch_size = 100\n",
    "\n",
    "history = vae.fit(x = x_train,\n",
    "                  y = x_train,\n",
    "                  validation_data = (x_train, x_train),\n",
    "                  epochs = epochs,\n",
    "                  batch_size = batch_size,\n",
    "                  verbose = 2)\n",
    "\n",
    "vae.save('cnn-vae_model.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 119,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(history.history['loss'])\n",
    "plt.plot(history.history['val_loss'])\n",
    "plt.title('Model loss')\n",
    "plt.ylabel('Loss')\n",
    "plt.xlabel('Epoch')\n",
    "plt.legend(['Train', 'Test'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 120,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['loss', 'kl_loss', 'recon_loss']"
      ]
     },
     "execution_count": 120,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "vae.metrics_names"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 121,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(history.history['kl_loss'])\n",
    "plt.title('Kullback-Liebler Loss')\n",
    "plt.ylabel('Loss')\n",
    "plt.xlabel('Epoch')\n",
    "plt.legend(['Train', 'Test'], loc = 'upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 122,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(history.history['recon_loss'])\n",
    "plt.title('Reconstruction (MSE) Loss')\n",
    "plt.ylabel('Loss')\n",
    "plt.xlabel('Epoch')\n",
    "plt.legend(['Train', 'Test'], loc = 'upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}