PredictTimeSeries-MultipleModelsByCorrelations-MSE1-Train-GRU.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 218,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from keras.models import Sequential\n",
    "from keras.layers import GRU\n",
    "from keras.layers import Dense\n",
    "from matplotlib import pyplot\n",
    "from sklearn.metrics import mean_absolute_error\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from keras.wrappers.scikit_learn import KerasClassifier\n",
    "import os\n",
    "\n",
    "def concat(dfs, concat_axis):\n",
    "    return pd.concat(dfs, axis=concat_axis)\n",
    "\n",
    "def repeatdata(start, times, df):\n",
    "    return df.loc[df.index.repeat(times)].reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 219,
   "metadata": {},
   "outputs": [],
   "source": [
    "shift = 4\n",
    "\n",
    "# Params\n",
    "data_directory = 'data'\n",
    "dataset_files = ['cpu_parameters', 'disk_io_parameters', 'jvm_parameters', 'memory_parameters', 'network_io_parameters']\n",
    "param_to_drop = []\n",
    "model_type = 'gru'\n",
    "model_epoches = 200\n",
    "model_batch_size = 200\n",
    "\n",
    "cols = ['@timestamp','system.cpu.user.pct', 'system.cpu.system.pct',\n",
    "       'system.cpu.idle.pct', 'system.cpu.iowait.pct',\n",
    "       'system.cpu.softirq.pct', 'system.cpu.total.pct',\n",
    "       'system.memory.used.pct',\n",
    "       'system.network.in.bytes', 'system.network.in.packets',\n",
    "       'system.network.in.dropped', 'system.network.out.bytes',\n",
    "       'system.network.out.packets', 'system.network.out.errors',\n",
    "       'system.diskio.iostat.await',\n",
    "       'system.diskio.iostat.queue.avg_size',\n",
    "       'system.diskio.iostat.read.per_sec.bytes',\n",
    "       'system.diskio.iostat.write.per_sec.bytes',\n",
    "       'jvm.metrics.memory.heap_memory_usage.committed',\n",
    "       'jvm.metrics.memory.heap_memory_usage.max',\n",
    "       'jvm.metrics.memory.heap_memory_usage.used',\n",
    "       'jvm.metrics.threading.thread_count',\n",
    "       'jvm.metrics.gc.psms.collection_count',\n",
    "       'jvm.metrics.gc.psms.collection_time',\n",
    "       'jvm.metrics.gc.pss.collection_count',\n",
    "       'jvm.metrics.gc.pss.collection_time', 'label']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 220,
   "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>@timestamp</th>\n",
       "      <th>system.cpu.user.pct</th>\n",
       "      <th>system.cpu.system.pct</th>\n",
       "      <th>system.cpu.idle.pct</th>\n",
       "      <th>system.cpu.iowait.pct</th>\n",
       "      <th>system.cpu.softirq.pct</th>\n",
       "      <th>system.cpu.total.pct</th>\n",
       "      <th>system.memory.used.pct</th>\n",
       "      <th>system.network.in.bytes</th>\n",
       "      <th>system.network.in.packets</th>\n",
       "      <th>...</th>\n",
       "      <th>system.diskio.iostat.write.per_sec.bytes</th>\n",
       "      <th>jvm.metrics.memory.heap_memory_usage.committed</th>\n",
       "      <th>jvm.metrics.memory.heap_memory_usage.max</th>\n",
       "      <th>jvm.metrics.memory.heap_memory_usage.used</th>\n",
       "      <th>jvm.metrics.threading.thread_count</th>\n",
       "      <th>jvm.metrics.gc.psms.collection_count</th>\n",
       "      <th>jvm.metrics.gc.psms.collection_time</th>\n",
       "      <th>jvm.metrics.gc.pss.collection_count</th>\n",
       "      <th>jvm.metrics.gc.pss.collection_time</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2018-01-01 00:00:00</td>\n",
       "      <td>0.7110</td>\n",
       "      <td>0.0675</td>\n",
       "      <td>3.1702</td>\n",
       "      <td>0.0302</td>\n",
       "      <td>0.0211</td>\n",
       "      <td>0.8298</td>\n",
       "      <td>0.8429</td>\n",
       "      <td>3000000.0</td>\n",
       "      <td>13038.0</td>\n",
       "      <td>...</td>\n",
       "      <td>50516.0</td>\n",
       "      <td>1.959264e+09</td>\n",
       "      <td>1.959264e+09</td>\n",
       "      <td>536266920.0</td>\n",
       "      <td>961.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>125</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2018-01-01 00:00:10</td>\n",
       "      <td>0.6695</td>\n",
       "      <td>0.0633</td>\n",
       "      <td>3.2410</td>\n",
       "      <td>0.0101</td>\n",
       "      <td>0.0161</td>\n",
       "      <td>0.7590</td>\n",
       "      <td>0.8471</td>\n",
       "      <td>3000000.0</td>\n",
       "      <td>13178.0</td>\n",
       "      <td>...</td>\n",
       "      <td>50516.0</td>\n",
       "      <td>1.959264e+09</td>\n",
       "      <td>1.959264e+09</td>\n",
       "      <td>536266920.0</td>\n",
       "      <td>961.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>125</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2018-01-01 00:00:20</td>\n",
       "      <td>0.5851</td>\n",
       "      <td>0.0755</td>\n",
       "      <td>3.3162</td>\n",
       "      <td>0.0020</td>\n",
       "      <td>0.0211</td>\n",
       "      <td>0.6838</td>\n",
       "      <td>0.8489</td>\n",
       "      <td>4000000.0</td>\n",
       "      <td>12945.0</td>\n",
       "      <td>...</td>\n",
       "      <td>50516.0</td>\n",
       "      <td>1.959264e+09</td>\n",
       "      <td>1.959264e+09</td>\n",
       "      <td>536266920.0</td>\n",
       "      <td>961.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>125</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2018-01-01 00:00:30</td>\n",
       "      <td>0.5350</td>\n",
       "      <td>0.0836</td>\n",
       "      <td>3.3552</td>\n",
       "      <td>0.0050</td>\n",
       "      <td>0.0212</td>\n",
       "      <td>0.6448</td>\n",
       "      <td>0.8547</td>\n",
       "      <td>3000000.0</td>\n",
       "      <td>13850.0</td>\n",
       "      <td>...</td>\n",
       "      <td>50516.0</td>\n",
       "      <td>1.959264e+09</td>\n",
       "      <td>1.959264e+09</td>\n",
       "      <td>536266920.0</td>\n",
       "      <td>961.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>125</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2018-01-01 00:00:40</td>\n",
       "      <td>0.6239</td>\n",
       "      <td>0.0584</td>\n",
       "      <td>3.2926</td>\n",
       "      <td>0.0040</td>\n",
       "      <td>0.0211</td>\n",
       "      <td>0.7074</td>\n",
       "      <td>0.8558</td>\n",
       "      <td>3000000.0</td>\n",
       "      <td>12838.0</td>\n",
       "      <td>...</td>\n",
       "      <td>50516.0</td>\n",
       "      <td>1.959264e+09</td>\n",
       "      <td>1.959264e+09</td>\n",
       "      <td>536266920.0</td>\n",
       "      <td>961.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>125</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 27 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            @timestamp  system.cpu.user.pct  system.cpu.system.pct  \\\n",
       "0  2018-01-01 00:00:00               0.7110                 0.0675   \n",
       "1  2018-01-01 00:00:10               0.6695                 0.0633   \n",
       "2  2018-01-01 00:00:20               0.5851                 0.0755   \n",
       "3  2018-01-01 00:00:30               0.5350                 0.0836   \n",
       "4  2018-01-01 00:00:40               0.6239                 0.0584   \n",
       "\n",
       "   system.cpu.idle.pct  system.cpu.iowait.pct  system.cpu.softirq.pct  \\\n",
       "0               3.1702                 0.0302                  0.0211   \n",
       "1               3.2410                 0.0101                  0.0161   \n",
       "2               3.3162                 0.0020                  0.0211   \n",
       "3               3.3552                 0.0050                  0.0212   \n",
       "4               3.2926                 0.0040                  0.0211   \n",
       "\n",
       "   system.cpu.total.pct  system.memory.used.pct  system.network.in.bytes  \\\n",
       "0                0.8298                  0.8429                3000000.0   \n",
       "1                0.7590                  0.8471                3000000.0   \n",
       "2                0.6838                  0.8489                4000000.0   \n",
       "3                0.6448                  0.8547                3000000.0   \n",
       "4                0.7074                  0.8558                3000000.0   \n",
       "\n",
       "   system.network.in.packets  ...    system.diskio.iostat.write.per_sec.bytes  \\\n",
       "0                    13038.0  ...                                     50516.0   \n",
       "1                    13178.0  ...                                     50516.0   \n",
       "2                    12945.0  ...                                     50516.0   \n",
       "3                    13850.0  ...                                     50516.0   \n",
       "4                    12838.0  ...                                     50516.0   \n",
       "\n",
       "   jvm.metrics.memory.heap_memory_usage.committed  \\\n",
       "0                                    1.959264e+09   \n",
       "1                                    1.959264e+09   \n",
       "2                                    1.959264e+09   \n",
       "3                                    1.959264e+09   \n",
       "4                                    1.959264e+09   \n",
       "\n",
       "   jvm.metrics.memory.heap_memory_usage.max  \\\n",
       "0                              1.959264e+09   \n",
       "1                              1.959264e+09   \n",
       "2                              1.959264e+09   \n",
       "3                              1.959264e+09   \n",
       "4                              1.959264e+09   \n",
       "\n",
       "   jvm.metrics.memory.heap_memory_usage.used  \\\n",
       "0                                536266920.0   \n",
       "1                                536266920.0   \n",
       "2                                536266920.0   \n",
       "3                                536266920.0   \n",
       "4                                536266920.0   \n",
       "\n",
       "   jvm.metrics.threading.thread_count  jvm.metrics.gc.psms.collection_count  \\\n",
       "0                               961.0                                     0   \n",
       "1                               961.0                                     0   \n",
       "2                               961.0                                     0   \n",
       "3                               961.0                                     0   \n",
       "4                               961.0                                     0   \n",
       "\n",
       "   jvm.metrics.gc.psms.collection_time  jvm.metrics.gc.pss.collection_count  \\\n",
       "0                                    0                                    2   \n",
       "1                                    0                                    2   \n",
       "2                                    0                                    2   \n",
       "3                                    0                                    2   \n",
       "4                                    0                                    2   \n",
       "\n",
       "   jvm.metrics.gc.pss.collection_time  label  \n",
       "0                                 125      0  \n",
       "1                                 125      0  \n",
       "2                                 125      0  \n",
       "3                                 125      0  \n",
       "4                                 125      0  \n",
       "\n",
       "[5 rows x 27 columns]"
      ]
     },
     "execution_count": 220,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cpu = pd.read_csv(data_directory+'/cpu_parameters.csv', header=0)\n",
    "memory = pd.read_csv(data_directory+'/memory_parameters.csv', header=0).drop([\"@timestamp\", \"label\"], axis=1)\n",
    "network = pd.read_csv(data_directory+'/network_io_parameters.csv', header=0).drop([\"@timestamp\", \"label\"], axis=1)\n",
    "diskio = repeatdata(0, 6, pd.read_csv(data_directory+'/disk_io_parameters.csv', header=0)).drop([\"@timestamp\", \"label\"], axis=1)\n",
    "jvm = repeatdata(0, 6, pd.read_csv(data_directory+'/jvm_parameters.csv', header=0)).drop([\"@timestamp\", \"label\"], axis=1)\n",
    "\n",
    "dataset = pd.concat([cpu, memory, network, diskio, jvm], 1)\n",
    "dataset = dataset[cols]\n",
    "dataset.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Set Multiple Model Parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 221,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_epoches = 200\n",
    "model_batch_size = 200"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Shifting the dataset\n",
    "shifting only one minute"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 222,
   "metadata": {},
   "outputs": [],
   "source": [
    "shift_param = shift*(-6)\n",
    "dataset_shifted = dataset.shift(shift_param)\n",
    "dataset = dataset.iloc[:shift_param]\n",
    "dataset_shifted = dataset_shifted.iloc[:shift_param]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Spliting the dataset \n",
    "Ratio - 80 : 20"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 223,
   "metadata": {},
   "outputs": [],
   "source": [
    "train = dataset.shape[0] * 2 // 10\n",
    "    \n",
    "train_dataset = dataset.iloc[:-train]\n",
    "train_dataset_shifted = dataset_shifted.iloc[:-train]\n",
    "\n",
    "test_dataset = dataset.iloc[-train:]\n",
    "test_dataset_shifted = dataset_shifted.iloc[-train:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 224,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/sudaraka/anaconda3/lib/python3.6/site-packages/ipykernel_launcher.py:2: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  \n",
      "/home/sudaraka/anaconda3/lib/python3.6/site-packages/ipykernel_launcher.py:3: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  This is separate from the ipykernel package so we can avoid doing imports until\n",
      "/home/sudaraka/anaconda3/lib/python3.6/site-packages/ipykernel_launcher.py:4: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  after removing the cwd from sys.path.\n"
     ]
    }
   ],
   "source": [
    "test_dataset.to_csv(data_directory+'/test_dataset.csv')\n",
    "test_dataset.drop([\"@timestamp\", \"label\"], axis=1, inplace=True)\n",
    "train_dataset.drop([\"@timestamp\", \"label\"], axis=1, inplace=True)\n",
    "train_dataset_shifted.drop([\"@timestamp\", \"label\"], axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Normalizing Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 225,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/sudaraka/anaconda3/lib/python3.6/site-packages/ipykernel_launcher.py:1: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  \"\"\"Entry point for launching an IPython kernel.\n",
      "/home/sudaraka/anaconda3/lib/python3.6/site-packages/pandas/core/indexing.py:537: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  self.obj[item] = s\n",
      "/home/sudaraka/anaconda3/lib/python3.6/site-packages/ipykernel_launcher.py:2: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  \n",
      "/home/sudaraka/anaconda3/lib/python3.6/site-packages/pandas/core/frame.py:2544: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  self.loc._setitem_with_indexer((slice(None), indexer), value)\n",
      "/home/sudaraka/anaconda3/lib/python3.6/site-packages/pandas/core/frame.py:2514: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  self._setitem_array(key, value)\n"
     ]
    }
   ],
   "source": [
    "train_dataset[train_dataset.columns] = StandardScaler().fit_transform(train_dataset)\n",
    "train_dataset_shifted[train_dataset_shifted.columns] = StandardScaler().fit_transform(train_dataset_shifted)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 226,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(57572, 25)\n"
     ]
    }
   ],
   "source": [
    "train_dataset = train_dataset.values.reshape((train_dataset.shape[0], 1, train_dataset.shape[1]))\n",
    "test_dataset_shaped = test_dataset.values.reshape((test_dataset.shape[0], 1, test_dataset.shape[1]))\n",
    "print(train_dataset_shifted.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Training Models"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Feature Set 1\n",
    "- system.cpu.iowait.pct'\n",
    "- system.diskio_sda.iostat.await'\n",
    "- system.diskio_sda.iostat.busy'\n",
    "- system.diskio_sda.iostat.queue.avg_size'\n",
    "- system.diskio_sda.iostat.read.request.merges_per_sec'\n",
    "- system.diskio_sda.iostat.read.request.per_sec'\n",
    "- system.diskio_sda.iostat.request.avg_size'\n",
    "- system.diskio_sda.iostat.service_time'\n",
    "- system.diskio_sda.iostat.write.request.merges_per_sec'\n",
    "- system.diskio_sda.iostat.write.request.per_sec'\n",
    "- system.diskio_sda2.iostat.await'\n",
    "- system.diskio_sda2.iostat.busy'\n",
    "- system.diskio_sda2.iostat.queue.avg_size'\n",
    "- system.diskio_sda2.iostat.read.request.merges_per_sec'\n",
    "- system.diskio_sda2.iostat.read.request.per_sec'\n",
    "- system.diskio_sda2.iostat.request.avg_size'\n",
    "- system.diskio_sda2.iostat.service_time'\n",
    "- system.diskio_sda2.iostat.write.request.merges_per_sec'\n",
    "- system.diskio_sda2.iostat.write.request.per_sec'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 227,
   "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>system.cpu.user.pct</th>\n",
       "      <th>system.cpu.system.pct</th>\n",
       "      <th>system.cpu.idle.pct</th>\n",
       "      <th>system.cpu.iowait.pct</th>\n",
       "      <th>system.cpu.softirq.pct</th>\n",
       "      <th>system.cpu.total.pct</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>-0.215992</td>\n",
       "      <td>0.134323</td>\n",
       "      <td>0.210056</td>\n",
       "      <td>-0.236885</td>\n",
       "      <td>-0.733666</td>\n",
       "      <td>-0.210056</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>-0.219927</td>\n",
       "      <td>-0.168956</td>\n",
       "      <td>0.246287</td>\n",
       "      <td>-0.121152</td>\n",
       "      <td>-0.705950</td>\n",
       "      <td>-0.246287</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-0.223487</td>\n",
       "      <td>-0.130272</td>\n",
       "      <td>0.247211</td>\n",
       "      <td>-0.236885</td>\n",
       "      <td>-0.456512</td>\n",
       "      <td>-0.247211</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>-0.169522</td>\n",
       "      <td>-0.034337</td>\n",
       "      <td>0.184177</td>\n",
       "      <td>-0.291995</td>\n",
       "      <td>-0.456512</td>\n",
       "      <td>-0.184177</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>-0.257777</td>\n",
       "      <td>-0.266438</td>\n",
       "      <td>0.302667</td>\n",
       "      <td>-0.291995</td>\n",
       "      <td>-1.010819</td>\n",
       "      <td>-0.302667</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   system.cpu.user.pct  system.cpu.system.pct  system.cpu.idle.pct  \\\n",
       "0            -0.215992               0.134323             0.210056   \n",
       "1            -0.219927              -0.168956             0.246287   \n",
       "2            -0.223487              -0.130272             0.247211   \n",
       "3            -0.169522              -0.034337             0.184177   \n",
       "4            -0.257777              -0.266438             0.302667   \n",
       "\n",
       "   system.cpu.iowait.pct  system.cpu.softirq.pct  system.cpu.total.pct  \n",
       "0              -0.236885               -0.733666             -0.210056  \n",
       "1              -0.121152               -0.705950             -0.246287  \n",
       "2              -0.236885               -0.456512             -0.247211  \n",
       "3              -0.291995               -0.456512             -0.184177  \n",
       "4              -0.291995               -1.010819             -0.302667  "
      ]
     },
     "execution_count": 227,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_dataset_shifted_1 = train_dataset_shifted[[\n",
    "    'system.cpu.user.pct',\n",
    "    'system.cpu.system.pct',\n",
    "    'system.cpu.idle.pct',\n",
    "    'system.cpu.iowait.pct',\n",
    "    'system.cpu.softirq.pct',\n",
    "    'system.cpu.total.pct'\n",
    "]]\n",
    "train_dataset_shifted_1.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 228,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/200\n",
      " - 11s - loss: 0.7694\n",
      "Epoch 2/200\n",
      " - 8s - loss: 0.5082\n",
      "Epoch 3/200\n",
      " - 7s - loss: 0.4011\n",
      "Epoch 4/200\n",
      " - 7s - loss: 0.3602\n",
      "Epoch 5/200\n",
      " - 7s - loss: 0.3434\n",
      "Epoch 6/200\n",
      " - 8s - loss: 0.3361\n",
      "Epoch 7/200\n",
      " - 9s - loss: 0.3331\n",
      "Epoch 8/200\n",
      " - 9s - loss: 0.3295\n",
      "Epoch 9/200\n",
      " - 8s - loss: 0.3279\n",
      "Epoch 10/200\n",
      " - 9s - loss: 0.3274\n",
      "Epoch 11/200\n",
      " - 9s - loss: 0.3272\n",
      "Epoch 12/200\n",
      " - 8s - loss: 0.3259\n",
      "Epoch 13/200\n",
      " - 7s - loss: 0.3258\n",
      "Epoch 14/200\n",
      " - 7s - loss: 0.3254\n",
      "Epoch 15/200\n",
      " - 7s - loss: 0.3253\n",
      "Epoch 16/200\n",
      " - 7s - loss: 0.3250\n",
      "Epoch 17/200\n",
      " - 7s - loss: 0.3249\n",
      "Epoch 18/200\n",
      " - 7s - loss: 0.3242\n",
      "Epoch 19/200\n",
      " - 7s - loss: 0.3242\n",
      "Epoch 20/200\n",
      " - 7s - loss: 0.3237\n",
      "Epoch 21/200\n",
      " - 7s - loss: 0.3234\n",
      "Epoch 22/200\n",
      " - 7s - loss: 0.3232\n",
      "Epoch 23/200\n",
      " - 7s - loss: 0.3229\n",
      "Epoch 24/200\n",
      " - 7s - loss: 0.3228\n",
      "Epoch 25/200\n",
      " - 7s - loss: 0.3226\n",
      "Epoch 26/200\n",
      " - 7s - loss: 0.3224\n",
      "Epoch 27/200\n",
      " - 8s - loss: 0.3222\n",
      "Epoch 28/200\n",
      " - 8s - loss: 0.3221\n",
      "Epoch 29/200\n",
      " - 8s - loss: 0.3219\n",
      "Epoch 30/200\n",
      " - 8s - loss: 0.3218\n",
      "Epoch 31/200\n",
      " - 8s - loss: 0.3216\n",
      "Epoch 32/200\n",
      " - 8s - loss: 0.3214\n",
      "Epoch 33/200\n",
      " - 8s - loss: 0.3212\n",
      "Epoch 34/200\n",
      " - 8s - loss: 0.3211\n",
      "Epoch 35/200\n",
      " - 9s - loss: 0.3209\n",
      "Epoch 36/200\n",
      " - 8s - loss: 0.3208\n",
      "Epoch 37/200\n",
      " - 9s - loss: 0.3205\n",
      "Epoch 38/200\n",
      " - 7s - loss: 0.3205\n",
      "Epoch 39/200\n",
      " - 7s - loss: 0.3202\n",
      "Epoch 40/200\n",
      " - 7s - loss: 0.3202\n",
      "Epoch 41/200\n",
      " - 7s - loss: 0.3199\n",
      "Epoch 42/200\n",
      " - 7s - loss: 0.3200\n",
      "Epoch 43/200\n",
      " - 7s - loss: 0.3196\n",
      "Epoch 44/200\n",
      " - 7s - loss: 0.3196\n",
      "Epoch 45/200\n",
      " - 7s - loss: 0.3194\n",
      "Epoch 46/200\n",
      " - 7s - loss: 0.3195\n",
      "Epoch 47/200\n",
      " - 8s - loss: 0.3192\n",
      "Epoch 48/200\n",
      " - 7s - loss: 0.3192\n",
      "Epoch 49/200\n",
      " - 7s - loss: 0.3191\n",
      "Epoch 50/200\n",
      " - 7s - loss: 0.3190\n",
      "Epoch 51/200\n",
      " - 7s - loss: 0.3188\n",
      "Epoch 52/200\n",
      " - 7s - loss: 0.3189\n",
      "Epoch 53/200\n",
      " - 7s - loss: 0.3187\n",
      "Epoch 54/200\n",
      " - 7s - loss: 0.3187\n",
      "Epoch 55/200\n",
      " - 7s - loss: 0.3185\n",
      "Epoch 56/200\n",
      " - 7s - loss: 0.3185\n",
      "Epoch 57/200\n",
      " - 7s - loss: 0.3183\n",
      "Epoch 58/200\n",
      " - 7s - loss: 0.3184\n",
      "Epoch 59/200\n",
      " - 7s - loss: 0.3182\n",
      "Epoch 60/200\n",
      " - 7s - loss: 0.3182\n",
      "Epoch 61/200\n",
      " - 7s - loss: 0.3183\n",
      "Epoch 62/200\n",
      " - 7s - loss: 0.3182\n",
      "Epoch 63/200\n",
      " - 7s - loss: 0.3183\n",
      "Epoch 64/200\n",
      " - 7s - loss: 0.3181\n",
      "Epoch 65/200\n",
      " - 7s - loss: 0.3182\n",
      "Epoch 66/200\n",
      " - 7s - loss: 0.3180\n",
      "Epoch 67/200\n",
      " - 7s - loss: 0.3182\n",
      "Epoch 68/200\n",
      " - 7s - loss: 0.3180\n",
      "Epoch 69/200\n",
      " - 7s - loss: 0.3181\n",
      "Epoch 70/200\n",
      " - 7s - loss: 0.3181\n",
      "Epoch 71/200\n",
      " - 7s - loss: 0.3181\n",
      "Epoch 72/200\n",
      " - 7s - loss: 0.3180\n",
      "Epoch 73/200\n",
      " - 7s - loss: 0.3180\n",
      "Epoch 74/200\n",
      " - 7s - loss: 0.3180\n",
      "Epoch 75/200\n",
      " - 7s - loss: 0.3180\n",
      "Epoch 76/200\n",
      " - 7s - loss: 0.3179\n",
      "Epoch 77/200\n",
      " - 7s - loss: 0.3179\n",
      "Epoch 78/200\n",
      " - 7s - loss: 0.3179\n",
      "Epoch 79/200\n",
      " - 7s - loss: 0.3178\n",
      "Epoch 80/200\n",
      " - 7s - loss: 0.3178\n",
      "Epoch 81/200\n",
      " - 7s - loss: 0.3177\n",
      "Epoch 82/200\n",
      " - 7s - loss: 0.3178\n",
      "Epoch 83/200\n",
      " - 7s - loss: 0.3177\n",
      "Epoch 84/200\n",
      " - 7s - loss: 0.3177\n",
      "Epoch 85/200\n",
      " - 7s - loss: 0.3177\n",
      "Epoch 86/200\n",
      " - 7s - loss: 0.3176\n",
      "Epoch 87/200\n",
      " - 7s - loss: 0.3176\n",
      "Epoch 88/200\n",
      " - 8s - loss: 0.3176\n",
      "Epoch 89/200\n",
      " - 7s - loss: 0.3176\n",
      "Epoch 90/200\n",
      " - 7s - loss: 0.3176\n",
      "Epoch 91/200\n",
      " - 7s - loss: 0.3176\n",
      "Epoch 92/200\n",
      " - 7s - loss: 0.3175\n",
      "Epoch 93/200\n",
      " - 7s - loss: 0.3175\n",
      "Epoch 94/200\n",
      " - 7s - loss: 0.3175\n",
      "Epoch 95/200\n",
      " - 7s - loss: 0.3175\n",
      "Epoch 96/200\n",
      " - 7s - loss: 0.3174\n",
      "Epoch 97/200\n",
      " - 7s - loss: 0.3174\n",
      "Epoch 98/200\n",
      " - 7s - loss: 0.3173\n",
      "Epoch 99/200\n",
      " - 7s - loss: 0.3174\n",
      "Epoch 100/200\n",
      " - 7s - loss: 0.3173\n",
      "Epoch 101/200\n",
      " - 7s - loss: 0.3174\n",
      "Epoch 102/200\n",
      " - 7s - loss: 0.3173\n",
      "Epoch 103/200\n",
      " - 7s - loss: 0.3174\n",
      "Epoch 104/200\n",
      " - 7s - loss: 0.3172\n",
      "Epoch 105/200\n",
      " - 7s - loss: 0.3173\n",
      "Epoch 106/200\n",
      " - 7s - loss: 0.3172\n",
      "Epoch 107/200\n",
      " - 7s - loss: 0.3173\n",
      "Epoch 108/200\n",
      " - 7s - loss: 0.3172\n",
      "Epoch 109/200\n",
      " - 8s - loss: 0.3172\n",
      "Epoch 110/200\n",
      " - 8s - loss: 0.3171\n",
      "Epoch 111/200\n",
      " - 7s - loss: 0.3172\n",
      "Epoch 112/200\n",
      " - 7s - loss: 0.3171\n",
      "Epoch 113/200\n",
      " - 7s - loss: 0.3171\n",
      "Epoch 114/200\n",
      " - 7s - loss: 0.3170\n",
      "Epoch 115/200\n",
      " - 7s - loss: 0.3171\n",
      "Epoch 116/200\n",
      " - 7s - loss: 0.3170\n",
      "Epoch 117/200\n",
      " - 7s - loss: 0.3171\n",
      "Epoch 118/200\n",
      " - 7s - loss: 0.3169\n",
      "Epoch 119/200\n",
      " - 7s - loss: 0.3170\n",
      "Epoch 120/200\n",
      " - 7s - loss: 0.3169\n",
      "Epoch 121/200\n",
      " - 7s - loss: 0.3170\n",
      "Epoch 122/200\n",
      " - 7s - loss: 0.3168\n",
      "Epoch 123/200\n",
      " - 7s - loss: 0.3169\n",
      "Epoch 124/200\n",
      " - 8s - loss: 0.3168\n",
      "Epoch 125/200\n",
      " - 7s - loss: 0.3169\n",
      "Epoch 126/200\n",
      " - 8s - loss: 0.3168\n",
      "Epoch 127/200\n",
      " - 7s - loss: 0.3169\n",
      "Epoch 128/200\n",
      " - 7s - loss: 0.3167\n",
      "Epoch 129/200\n",
      " - 7s - loss: 0.3168\n",
      "Epoch 130/200\n",
      " - 7s - loss: 0.3167\n",
      "Epoch 131/200\n",
      " - 7s - loss: 0.3167\n",
      "Epoch 132/200\n",
      " - 7s - loss: 0.3166\n",
      "Epoch 133/200\n",
      " - 7s - loss: 0.3167\n",
      "Epoch 134/200\n",
      " - 7s - loss: 0.3165\n",
      "Epoch 135/200\n",
      " - 7s - loss: 0.3166\n",
      "Epoch 136/200\n",
      " - 7s - loss: 0.3165\n",
      "Epoch 137/200\n",
      " - 7s - loss: 0.3166\n",
      "Epoch 138/200\n",
      " - 7s - loss: 0.3165\n",
      "Epoch 139/200\n",
      " - 7s - loss: 0.3166\n",
      "Epoch 140/200\n",
      " - 7s - loss: 0.3164\n",
      "Epoch 141/200\n",
      " - 7s - loss: 0.3165\n",
      "Epoch 142/200\n",
      " - 7s - loss: 0.3165\n",
      "Epoch 143/200\n",
      " - 7s - loss: 0.3165\n",
      "Epoch 144/200\n",
      " - 7s - loss: 0.3163\n",
      "Epoch 145/200\n",
      " - 7s - loss: 0.3165\n",
      "Epoch 146/200\n",
      " - 7s - loss: 0.3163\n",
      "Epoch 147/200\n",
      " - 7s - loss: 0.3164\n",
      "Epoch 148/200\n",
      " - 7s - loss: 0.3162\n",
      "Epoch 149/200\n",
      " - 7s - loss: 0.3163\n",
      "Epoch 150/200\n",
      " - 7s - loss: 0.3162\n",
      "Epoch 151/200\n",
      " - 7s - loss: 0.3163\n",
      "Epoch 152/200\n",
      " - 7s - loss: 0.3162\n",
      "Epoch 153/200\n",
      " - 7s - loss: 0.3162\n",
      "Epoch 154/200\n",
      " - 7s - loss: 0.3161\n",
      "Epoch 155/200\n",
      " - 7s - loss: 0.3162\n",
      "Epoch 156/200\n",
      " - 7s - loss: 0.3161\n",
      "Epoch 157/200\n",
      " - 7s - loss: 0.3161\n",
      "Epoch 158/200\n",
      " - 7s - loss: 0.3160\n",
      "Epoch 159/200\n",
      " - 7s - loss: 0.3161\n",
      "Epoch 160/200\n",
      " - 7s - loss: 0.3160\n",
      "Epoch 161/200\n",
      " - 7s - loss: 0.3160\n",
      "Epoch 162/200\n",
      " - 7s - loss: 0.3159\n",
      "Epoch 163/200\n",
      " - 7s - loss: 0.3160\n",
      "Epoch 164/200\n",
      " - 7s - loss: 0.3161\n",
      "Epoch 165/200\n",
      " - 7s - loss: 0.3159\n",
      "Epoch 166/200\n",
      " - 7s - loss: 0.3159\n",
      "Epoch 167/200\n",
      " - 7s - loss: 0.3158\n",
      "Epoch 168/200\n",
      " - 7s - loss: 0.3158\n",
      "Epoch 169/200\n",
      " - 7s - loss: 0.3158\n",
      "Epoch 170/200\n",
      " - 8s - loss: 0.3159\n",
      "Epoch 171/200\n",
      " - 9s - loss: 0.3158\n",
      "Epoch 172/200\n",
      " - 8s - loss: 0.3157\n",
      "Epoch 173/200\n",
      " - 8s - loss: 0.3157\n",
      "Epoch 174/200\n",
      " - 8s - loss: 0.3157\n",
      "Epoch 175/200\n",
      " - 8s - loss: 0.3157\n",
      "Epoch 176/200\n",
      " - 8s - loss: 0.3158\n",
      "Epoch 177/200\n",
      " - 8s - loss: 0.3156\n",
      "Epoch 178/200\n",
      " - 8s - loss: 0.3156\n",
      "Epoch 179/200\n",
      " - 8s - loss: 0.3156\n",
      "Epoch 180/200\n",
      " - 8s - loss: 0.3155\n",
      "Epoch 181/200\n",
      " - 8s - loss: 0.3155\n",
      "Epoch 182/200\n",
      " - 8s - loss: 0.3157\n",
      "Epoch 183/200\n",
      " - 8s - loss: 0.3155\n",
      "Epoch 184/200\n",
      " - 8s - loss: 0.3155\n",
      "Epoch 185/200\n",
      " - 8s - loss: 0.3154\n",
      "Epoch 186/200\n",
      " - 8s - loss: 0.3154\n",
      "Epoch 187/200\n",
      " - 8s - loss: 0.3154\n",
      "Epoch 188/200\n",
      " - 8s - loss: 0.3154\n",
      "Epoch 189/200\n",
      " - 9s - loss: 0.3153\n",
      "Epoch 190/200\n",
      " - 9s - loss: 0.3153\n",
      "Epoch 191/200\n",
      " - 8s - loss: 0.3153\n",
      "Epoch 192/200\n",
      " - 8s - loss: 0.3152\n",
      "Epoch 193/200\n",
      " - 8s - loss: 0.3152\n",
      "Epoch 194/200\n",
      " - 8s - loss: 0.3153\n",
      "Epoch 195/200\n",
      " - 8s - loss: 0.3151\n",
      "Epoch 196/200\n",
      " - 8s - loss: 0.3152\n",
      "Epoch 197/200\n",
      " - 8s - loss: 0.3152\n",
      "Epoch 198/200\n",
      " - 8s - loss: 0.3151\n",
      "Epoch 199/200\n",
      " - 8s - loss: 0.3151\n",
      "Epoch 200/200\n",
      " - 8s - loss: 0.3151\n"
     ]
    }
   ],
   "source": [
    "model_lstm_1 = Sequential()\n",
    "model_lstm_1.add(GRU(10, input_shape=(1, train_dataset_shifted.shape[1])))\n",
    "model_lstm_1.add(Dense(6))\n",
    "model_lstm_1.compile(loss='mse', optimizer='adam')\n",
    "\n",
    "# history = model_lstm_1.fit(train_dataset, train_dataset_shifted_1, verbose=2, shuffle=False, epochs=model_epoches, validation_data=(test_dataset_shaped ,test_dataset_shifted_1), batch_size= 200)\n",
    "history = model_lstm_1.fit(train_dataset, train_dataset_shifted_1, verbose=2, shuffle=False, epochs=model_epoches)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Feature Set 2\n",
    "- system.cpu.total.pct'\n",
    "- system.cpu.user.pct'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "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>system.network.in.bytes</th>\n",
       "      <th>system.network.out.bytes</th>\n",
       "      <th>system.network.in.dropped</th>\n",
       "      <th>system.network.out.errors</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.242874</td>\n",
       "      <td>0.432520</td>\n",
       "      <td>0.192984</td>\n",
       "      <td>0.265885</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.242874</td>\n",
       "      <td>0.432520</td>\n",
       "      <td>0.192984</td>\n",
       "      <td>0.651874</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.242874</td>\n",
       "      <td>0.432520</td>\n",
       "      <td>0.192984</td>\n",
       "      <td>-0.506092</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.242874</td>\n",
       "      <td>-0.907109</td>\n",
       "      <td>-0.999641</td>\n",
       "      <td>1.809841</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.242874</td>\n",
       "      <td>0.432520</td>\n",
       "      <td>-0.999641</td>\n",
       "      <td>-0.120104</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   system.network.in.bytes  system.network.out.bytes  \\\n",
       "0                 0.242874                  0.432520   \n",
       "1                 0.242874                  0.432520   \n",
       "2                 0.242874                  0.432520   \n",
       "3                 0.242874                 -0.907109   \n",
       "4                 0.242874                  0.432520   \n",
       "\n",
       "   system.network.in.dropped  system.network.out.errors  \n",
       "0                   0.192984                   0.265885  \n",
       "1                   0.192984                   0.651874  \n",
       "2                   0.192984                  -0.506092  \n",
       "3                  -0.999641                   1.809841  \n",
       "4                  -0.999641                  -0.120104  "
      ]
     },
     "execution_count": 229,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_dataset_shifted_2 = train_dataset_shifted[[\n",
    "    'system.network.in.bytes',\n",
    "    'system.network.out.bytes',\n",
    "    'system.network.in.dropped',\n",
    "    'system.network.out.errors'  \n",
    "]]\n",
    "\n",
    "train_dataset_shifted_2.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/200\n",
      " - 12s - loss: 0.9172\n",
      "Epoch 2/200\n",
      " - 9s - loss: 0.6865\n",
      "Epoch 3/200\n",
      " - 9s - loss: 0.6076\n",
      "Epoch 4/200\n",
      " - 8s - loss: 0.5869\n",
      "Epoch 5/200\n",
      " - 7s - loss: 0.5805\n",
      "Epoch 6/200\n",
      " - 8s - loss: 0.5773\n",
      "Epoch 7/200\n",
      " - 8s - loss: 0.5753\n",
      "Epoch 8/200\n",
      " - 7s - loss: 0.5740\n",
      "Epoch 9/200\n",
      " - 8s - loss: 0.5730\n",
      "Epoch 10/200\n",
      " - 7s - loss: 0.5722\n",
      "Epoch 11/200\n",
      " - 7s - loss: 0.5715\n",
      "Epoch 12/200\n",
      " - 7s - loss: 0.5709\n",
      "Epoch 13/200\n",
      " - 8s - loss: 0.5704\n",
      "Epoch 14/200\n",
      " - 8s - loss: 0.5698\n",
      "Epoch 15/200\n",
      " - 7s - loss: 0.5694\n",
      "Epoch 16/200\n",
      " - 7s - loss: 0.5690\n",
      "Epoch 17/200\n",
      " - 7s - loss: 0.5687\n",
      "Epoch 18/200\n",
      " - 8s - loss: 0.5682\n",
      "Epoch 19/200\n",
      " - 7s - loss: 0.5679\n",
      "Epoch 20/200\n",
      " - 7s - loss: 0.5675\n",
      "Epoch 21/200\n",
      " - 8s - loss: 0.5671\n",
      "Epoch 22/200\n",
      " - 8s - loss: 0.5667\n",
      "Epoch 23/200\n",
      " - 7s - loss: 0.5664\n",
      "Epoch 24/200\n",
      " - 8s - loss: 0.5661\n",
      "Epoch 25/200\n",
      " - 9s - loss: 0.5658\n",
      "Epoch 26/200\n",
      " - 9s - loss: 0.5655\n",
      "Epoch 27/200\n",
      " - 9s - loss: 0.5652\n",
      "Epoch 28/200\n",
      " - 9s - loss: 0.5649\n",
      "Epoch 29/200\n",
      " - 9s - loss: 0.5646\n",
      "Epoch 30/200\n",
      " - 9s - loss: 0.5644\n",
      "Epoch 31/200\n",
      " - 9s - loss: 0.5641\n",
      "Epoch 32/200\n",
      " - 9s - loss: 0.5638\n",
      "Epoch 33/200\n",
      " - 9s - loss: 0.5636\n",
      "Epoch 34/200\n",
      " - 9s - loss: 0.5634\n",
      "Epoch 35/200\n",
      " - 9s - loss: 0.5632\n",
      "Epoch 36/200\n",
      " - 9s - loss: 0.5630\n",
      "Epoch 37/200\n",
      " - 9s - loss: 0.5628\n",
      "Epoch 38/200\n",
      " - 9s - loss: 0.5626\n",
      "Epoch 39/200\n",
      " - 9s - loss: 0.5624\n",
      "Epoch 40/200\n",
      " - 9s - loss: 0.5622\n",
      "Epoch 41/200\n",
      " - 9s - loss: 0.5620\n",
      "Epoch 42/200\n",
      " - 9s - loss: 0.5619\n",
      "Epoch 43/200\n",
      " - 7s - loss: 0.5617\n",
      "Epoch 44/200\n"
     ]
    }
   ],
   "source": [
    "model_lstm_2 = Sequential()\n",
    "# model_lstm_2.add(GRU(10, input_shape=(1, train_dataset_shifted.shape[1]), activation='softplus', recurrent_activation='linear'))\n",
    "model_lstm_2.add(GRU(10, input_shape=(1, train_dataset_shifted.shape[1])))\n",
    "model_lstm_2.add(Dense(4))\n",
    "model_lstm_2.compile(loss='mse', optimizer='adam')\n",
    "\n",
    "model_lstm_2.fit(train_dataset, train_dataset_shifted_2, verbose=2, shuffle=False, epochs=model_epoches)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Feature Set 3\n",
    "- jolokia.metrics.memory.heap_memory_usage.committed'\n",
    "- jolokia.metrics.memory.heap_memory_usage.max'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_dataset_shifted_3 = train_dataset_shifted[[\n",
    "    'system.network.in.packets',\n",
    "    'system.network.out.packets'\n",
    "]]\n",
    "\n",
    "train_dataset_shifted_3.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_lstm_3 = Sequential()\n",
    "# model_lstm_3.add(GRU(10, input_shape=(1, train_dataset_shifted.shape[1]), activation='softplus', recurrent_activation='linear'))\n",
    "model_lstm_3.add(GRU(10, input_shape=(1, train_dataset_shifted.shape[1])))\n",
    "model_lstm_3.add(Dense(2))\n",
    "model_lstm_3.compile(loss='mse', optimizer='adam')\n",
    "\n",
    "model_lstm_3.fit(train_dataset, train_dataset_shifted_3, verbose=2, shuffle=False, epochs=model_epoches)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Feature Set 4\n",
    "- system.load.1'\n",
    "- system.load.15'\n",
    "- system.load.5'\n",
    "- system.load.norm.1'\n",
    "- system.load.norm.15'\n",
    "- system.load.norm.5'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_dataset_shifted_4 = train_dataset_shifted[[\n",
    "    'system.diskio.iostat.await',\n",
    "    'system.diskio.iostat.queue.avg_size',\n",
    "    'system.diskio.iostat.read.per_sec.bytes',\n",
    "    'system.diskio.iostat.write.per_sec.bytes'\n",
    "]]\n",
    "\n",
    "train_dataset_shifted_4.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "model_lstm_4 = Sequential()\n",
    "# model_lstm_4.add(GRU(10, input_shape=(1, train_dataset_shifted.shape[1]), activation='softplus', recurrent_activation='linear'))\n",
    "model_lstm_4.add(GRU(10, input_shape=(1, train_dataset_shifted.shape[1])))\n",
    "model_lstm_4.add(Dense(4))\n",
    "model_lstm_4.compile(loss='mse', optimizer='adam')\n",
    "\n",
    "model_lstm_4.fit(train_dataset, train_dataset_shifted_4, verbose=2, shuffle=False, epochs=model_epoches)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Feature Set 5\n",
    "- system.cpu.idle.pct'\n",
    "- system.cpu.softirq.pct'\n",
    "- system.cpu.system.pct'\n",
    "- jolokia.metrics.memory.heap_memory_usage.used'\n",
    "- jolokia.metrics.memory.non_heap_memory_usage.used'\n",
    "- jolokia.metrics.threading.daemon_thread_count'\n",
    "- jolokia.metrics.threading.thread_count'\n",
    "- system.memory.actual.used.pct'\n",
    "- system.memory.swap.used.pct'\n",
    "- system.memory.used.pct'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_dataset_shifted_5 = train_dataset_shifted[[\n",
    "    'jvm.metrics.memory.heap_memory_usage.committed',\n",
    "    'jvm.metrics.memory.heap_memory_usage.max'\n",
    "]]\n",
    "\n",
    "train_dataset_shifted_5.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "model_lstm_5 = Sequential()\n",
    "# model_lstm_5.add(GRU(10, input_shape=(1, train_dataset_shifted.shape[1]), activation='softplus', recurrent_activation='linear'))\n",
    "model_lstm_5.add(GRU(10, input_shape=(1, train_dataset_shifted.shape[1])))\n",
    "model_lstm_5.add(Dense(2))\n",
    "model_lstm_5.compile(loss='mse', optimizer='adam')\n",
    "\n",
    "model_lstm_5.fit(train_dataset, train_dataset_shifted_5, verbose=2, shuffle=False, epochs=model_epoches)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_dataset_shifted_6 = train_dataset_shifted[[\n",
    "    'jvm.metrics.memory.heap_memory_usage.used',\n",
    "    'jvm.metrics.gc.psms.collection_count',\n",
    "    'jvm.metrics.gc.psms.collection_time',\n",
    "    'system.memory.used.pct']]\n",
    "\n",
    "train_dataset_shifted_6.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "model_lstm_6 = Sequential()\n",
    "# model_lstm_6.add(GRU(10, input_shape=(1, train_dataset_shifted.shape[1]), activation='softplus', recurrent_activation='linear'))\n",
    "model_lstm_6.add(GRU(10, input_shape=(1, train_dataset_shifted.shape[1])))\n",
    "model_lstm_6.add(Dense(4))\n",
    "model_lstm_6.compile(loss='mse', optimizer='adam')\n",
    "\n",
    "model_lstm_6.fit(train_dataset, train_dataset_shifted_6, verbose=2, shuffle=False, epochs=model_epoches)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_dataset_shifted_7 = train_dataset_shifted[[\n",
    "    'jvm.metrics.gc.pss.collection_count',\n",
    "    'jvm.metrics.gc.pss.collection_time',\n",
    "    'jvm.metrics.threading.thread_count'\n",
    "]]\n",
    "\n",
    "train_dataset_shifted_7.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "model_lstm_7 = Sequential()\n",
    "# model_lstm_7.add(GRU(10, input_shape=(1, train_dataset_shifted.shape[1]), activation='softplus', recurrent_activation='linear'))\n",
    "model_lstm_7.add(GRU(10, input_shape=(1, train_dataset_shifted.shape[1])))\n",
    "model_lstm_7.add(Dense(3))\n",
    "model_lstm_7.compile(loss='mse', optimizer='adam')\n",
    "\n",
    "model_lstm_7.fit(train_dataset, train_dataset_shifted_7, verbose=2, shuffle=False, epochs=model_epoches)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Serializing Models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "directory = data_directory+\"/models/shift-\"+str(shift)\n",
    "models = [model_lstm_1, model_lstm_2, model_lstm_3, model_lstm_4, model_lstm_5, model_lstm_6, model_lstm_7]\n",
    "\n",
    "if not os.path.exists(directory):\n",
    "    os.makedirs(directory)\n",
    "\n",
    "count = 0\n",
    "for model_to_save in models:\n",
    "    count += 1\n",
    "    model_json = model_to_save.to_json()\n",
    "    with open(directory+\"/model_\"+ model_type +\"_\"+str(count)+\".json\", \"w\") as json_file:\n",
    "        json_file.write(model_json)\n",
    "\n",
    "        # serialize weights to HDF5\n",
    "    model_to_save.save_weights(directory+\"/model_gru_\"+str(count)+\".h5\")\n",
    "\n",
    "print(\"Saved models to disk\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}