HARAccGyroData.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from scipy import stats\n",
    "from sklearn import metrics\n",
    "from sklearn.model_selection import train_test_split\n",
    "from keras.models import Sequential, Model\n",
    "from keras.layers import Dense\n",
    "from keras.layers import Flatten\n",
    "from keras.layers import Dropout\n",
    "from keras.callbacks import ModelCheckpoint\n",
    "from keras.layers.convolutional import Conv1D\n",
    "from keras.layers.convolutional import MaxPooling1D\n",
    "from sklearn.metrics import classification_report\n",
    "RANDOM_SEED = 42\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['P1', 'P2', 'P3', 'P4', 'P5', 'P6']\n"
     ]
    }
   ],
   "source": [
    "dataset_path = './Dataset/'\n",
    "person_folders = os.listdir(dataset_path)\n",
    "print(person_folders)\n",
    "\n",
    "df = pd.DataFrame()\n",
    "df_list = []\n",
    "\n",
    "for i in range(len(person_folders)):\n",
    "    person_folder = person_folders[i]\n",
    "    csv_file_path = dataset_path + person_folder + '/'\n",
    "    csv_files = os.listdir(csv_file_path)\n",
    "\n",
    "    for j in range(len(csv_files)):\n",
    "        csv_file = csv_files[j]\n",
    "        #print(csv_file)\n",
    "\n",
    "        data_frame = pd.read_csv(csv_file_path + '/' + csv_file, index_col=None, header=None)\n",
    "        df_list.append(data_frame)\n",
    "\n",
    "df = pd.concat(df_list, axis = 0, sort= True, ignore_index = True)\n",
    "df.columns = ['Timestamp', 'Ax', 'Ay', 'Az', 'Gx', 'Gy', 'Gz', 'Label', 'Activity']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Timestamp    105881\n",
      "Ax           105881\n",
      "Ay           105881\n",
      "Az           105881\n",
      "Gx           105881\n",
      "Gy           105881\n",
      "Gz           105881\n",
      "Label        105881\n",
      "Activity     105881\n",
      "dtype: int64\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Timestamp</th>\n",
       "      <th>Ax</th>\n",
       "      <th>Ay</th>\n",
       "      <th>Az</th>\n",
       "      <th>Gx</th>\n",
       "      <th>Gy</th>\n",
       "      <th>Gz</th>\n",
       "      <th>Label</th>\n",
       "      <th>Activity</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1.564040e+12</td>\n",
       "      <td>0.499756</td>\n",
       "      <td>0.798584</td>\n",
       "      <td>-0.313110</td>\n",
       "      <td>-16.768293</td>\n",
       "      <td>-10.091463</td>\n",
       "      <td>11.189024</td>\n",
       "      <td>4</td>\n",
       "      <td>Eating</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1.564040e+12</td>\n",
       "      <td>0.497314</td>\n",
       "      <td>0.826172</td>\n",
       "      <td>-0.312622</td>\n",
       "      <td>-6.981707</td>\n",
       "      <td>-11.310976</td>\n",
       "      <td>10.731707</td>\n",
       "      <td>4</td>\n",
       "      <td>Eating</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1.564040e+12</td>\n",
       "      <td>0.495605</td>\n",
       "      <td>0.825562</td>\n",
       "      <td>-0.313477</td>\n",
       "      <td>-5.823171</td>\n",
       "      <td>-11.402439</td>\n",
       "      <td>3.993902</td>\n",
       "      <td>4</td>\n",
       "      <td>Eating</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1.564040e+12</td>\n",
       "      <td>0.493408</td>\n",
       "      <td>0.823242</td>\n",
       "      <td>-0.314697</td>\n",
       "      <td>-5.091463</td>\n",
       "      <td>-11.067073</td>\n",
       "      <td>6.890244</td>\n",
       "      <td>4</td>\n",
       "      <td>Eating</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1.564040e+12</td>\n",
       "      <td>0.495605</td>\n",
       "      <td>0.819336</td>\n",
       "      <td>-0.315552</td>\n",
       "      <td>-4.115854</td>\n",
       "      <td>-11.189024</td>\n",
       "      <td>16.890244</td>\n",
       "      <td>4</td>\n",
       "      <td>Eating</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      Timestamp        Ax        Ay        Az         Gx         Gy  \\\n",
       "0  1.564040e+12  0.499756  0.798584 -0.313110 -16.768293 -10.091463   \n",
       "1  1.564040e+12  0.497314  0.826172 -0.312622  -6.981707 -11.310976   \n",
       "2  1.564040e+12  0.495605  0.825562 -0.313477  -5.823171 -11.402439   \n",
       "3  1.564040e+12  0.493408  0.823242 -0.314697  -5.091463 -11.067073   \n",
       "4  1.564040e+12  0.495605  0.819336 -0.315552  -4.115854 -11.189024   \n",
       "\n",
       "          Gz  Label Activity  \n",
       "0  11.189024      4   Eating  \n",
       "1  10.731707      4   Eating  \n",
       "2   3.993902      4   Eating  \n",
       "3   6.890244      4   Eating  \n",
       "4  16.890244      4   Eating  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(df.count())\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df['Activity'].value_counts().plot(kind='bar', title='Records by Activity Type', figsize=(8, 6),align='center');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Normalize features for training data set (values between 0 and 1)\n",
    "# Surpress warning for next 3 operation\n",
    "pd.options.mode.chained_assignment = None  # default='warn'\n",
    "\n",
    "df['Ax'] = df['Ax'] / df['Ax'].max()\n",
    "df['Ay'] = df['Ay'] / df['Ay'].max()\n",
    "df['Az'] = df['Az'] / df['Az'].max()\n",
    "\n",
    "df['Gx'] = df['Gx'] / df['Gx'].max()\n",
    "df['Gy'] = df['Gy'] / df['Gy'].max()\n",
    "df['Gz'] = df['Gz'] / df['Gz'].max()\n",
    "\n",
    "# Round numbers\n",
    "df = df.round({'Ax': 5, 'Ay': 5, 'Az': 5, 'Gx': 5, 'Gy': 5, 'Gz': 5})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\admin\\Anaconda3\\lib\\site-packages\\scipy\\stats\\stats.py:245: RuntimeWarning: The input array could not be properly checked for nan values. nan values will be ignored.\n",
      "  \"values. nan values will be ignored.\", RuntimeWarning)\n"
     ]
    }
   ],
   "source": [
    "N_TIME_STEPS = 80\n",
    "N_FEATURES = 6\n",
    "step = 20\n",
    "segments = []\n",
    "labels = []\n",
    "for i in range(0, len(df) - N_TIME_STEPS, step):\n",
    "    Ax = df['Ax'].values[i: i + N_TIME_STEPS]\n",
    "    Ay = df['Ay'].values[i: i + N_TIME_STEPS]\n",
    "    Az = df['Az'].values[i: i + N_TIME_STEPS]\n",
    "    \n",
    "    Gx = df['Gx'].values[i: i + N_TIME_STEPS]\n",
    "    Gy = df['Gy'].values[i: i + N_TIME_STEPS]\n",
    "    Gz = df['Gz'].values[i: i + N_TIME_STEPS]\n",
    "    \n",
    "    label = stats.mode(df['Activity'][i: i + N_TIME_STEPS])[0][0]\n",
    "    segments.append([Ax, Ay, Az, Gx, Gy, Gz])\n",
    "    labels.append(label)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "reshaped_segments = np.asarray(segments, dtype= np.float32).reshape(-1, N_TIME_STEPS, N_FEATURES)\n",
    "labels = np.asarray(pd.get_dummies(labels), dtype = np.float32)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(\n",
    "        reshaped_segments, labels, test_size=0.2, random_state=RANDOM_SEED)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_train :  (4232, 80, 6)\n",
      "X_test :  (1059, 80, 6)\n"
     ]
    }
   ],
   "source": [
    "print('X_train : ', X_train.shape)\n",
    "print('X_test : ', X_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "n_timesteps :  80\n",
      "n_features :  6\n",
      "n_outputs :  7\n"
     ]
    }
   ],
   "source": [
    "verbose = 1\n",
    "epochs = 50\n",
    "batch_size = 100\n",
    "\n",
    "n_timesteps = X_train.shape[1]\n",
    "n_features = X_train.shape[2]\n",
    "n_outputs = y_train.shape[1]\n",
    "\n",
    "print('n_timesteps : ', n_timesteps)\n",
    "print('n_features : ', n_features)\n",
    "print('n_outputs : ', n_outputs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = Sequential()\n",
    "model.add(Conv1D(filters=64, kernel_size=3, activation='relu', input_shape=(n_timesteps,n_features)))\n",
    "model.add(Conv1D(filters=64, kernel_size=3, activation='relu', padding = 'same'))\n",
    "model.add(Dropout(0.5))\n",
    "model.add(MaxPooling1D(pool_size=2))\n",
    "model.add(Flatten())\n",
    "model.add(Dense(100, activation='relu'))\n",
    "model.add(Dense(n_outputs, activation='softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 4232 samples, validate on 1059 samples\n",
      "Epoch 1/50\n",
      "4232/4232 [==============================] - 3s 788us/step - loss: 1.0188 - acc: 0.6151 - val_loss: 0.5761 - val_acc: 0.8149: 1.24\n",
      "Epoch 2/50\n",
      "4232/4232 [==============================] - 3s 657us/step - loss: 0.4747 - acc: 0.8228 - val_loss: 0.4950 - val_acc: 0.8234\n",
      "Epoch 3/50\n",
      "4232/4232 [==============================] - 3s 655us/step - loss: 0.3922 - acc: 0.8495 - val_loss: 0.3923 - val_acc: 0.8669\n",
      "Epoch 4/50\n",
      "4232/4232 [==============================] - 3s 708us/step - loss: 0.3686 - acc: 0.8518 - val_loss: 0.3990 - val_acc: 0.8716\n",
      "Epoch 5/50\n",
      "4232/4232 [==============================] - 3s 753us/step - loss: 0.3236 - acc: 0.8771 - val_loss: 0.3413 - val_acc: 0.8829\n",
      "Epoch 6/50\n",
      "4232/4232 [==============================] - 3s 651us/step - loss: 0.2944 - acc: 0.8816 - val_loss: 0.3393 - val_acc: 0.87350.2959 - acc\n",
      "Epoch 7/50\n",
      "4232/4232 [==============================] - 3s 658us/step - loss: 0.2727 - acc: 0.8913 - val_loss: 0.3128 - val_acc: 0.8791\n",
      "Epoch 8/50\n",
      "4232/4232 [==============================] - 3s 643us/step - loss: 0.2695 - acc: 0.8904 - val_loss: 0.3048 - val_acc: 0.8820\n",
      "Epoch 9/50\n",
      "4232/4232 [==============================] - 3s 650us/step - loss: 0.2357 - acc: 0.9062 - val_loss: 0.2937 - val_acc: 0.8933oss: 0.2358 - acc: \n",
      "Epoch 10/50\n",
      "4232/4232 [==============================] - 3s 753us/step - loss: 0.2286 - acc: 0.9126 - val_loss: 0.2956 - val_acc: 0.8725\n",
      "Epoch 11/50\n",
      "4232/4232 [==============================] - 3s 665us/step - loss: 0.2204 - acc: 0.9119 - val_loss: 0.2428 - val_acc: 0.9131\n",
      "Epoch 12/50\n",
      "4232/4232 [==============================] - 3s 645us/step - loss: 0.1890 - acc: 0.9272 - val_loss: 0.2574 - val_acc: 0.9018\n",
      "Epoch 13/50\n",
      "4232/4232 [==============================] - 3s 678us/step - loss: 0.1803 - acc: 0.9293 - val_loss: 0.2632 - val_acc: 0.8952\n",
      "Epoch 14/50\n",
      "4232/4232 [==============================] - 3s 647us/step - loss: 0.1815 - acc: 0.9260 - val_loss: 0.2310 - val_acc: 0.9292\n",
      "Epoch 15/50\n",
      "4232/4232 [==============================] - 3s 752us/step - loss: 0.1592 - acc: 0.9388 - val_loss: 0.2515 - val_acc: 0.8961- loss: 0.1499 - a - ETA: 1s - loss: 0.1513 -  - ETA: 0s - loss: 0.1546 -\n",
      "Epoch 16/50\n",
      "4232/4232 [==============================] - 3s 786us/step - loss: 0.1617 - acc: 0.9379 - val_loss: 0.2422 - val_acc: 0.9027\n",
      "Epoch 17/50\n",
      "4232/4232 [==============================] - 3s 677us/step - loss: 0.1421 - acc: 0.9459 - val_loss: 0.2368 - val_acc: 0.9207\n",
      "Epoch 18/50\n",
      "4232/4232 [==============================] - 3s 666us/step - loss: 0.1506 - acc: 0.9390 - val_loss: 0.2243 - val_acc: 0.9235\n",
      "Epoch 19/50\n",
      "4232/4232 [==============================] - 3s 662us/step - loss: 0.1335 - acc: 0.9480 - val_loss: 0.2161 - val_acc: 0.9197\n",
      "Epoch 20/50\n",
      "4232/4232 [==============================] - 3s 690us/step - loss: 0.1249 - acc: 0.9499 - val_loss: 0.2638 - val_acc: 0.9084\n",
      "Epoch 21/50\n",
      "4232/4232 [==============================] - 4s 851us/step - loss: 0.1299 - acc: 0.9513 - val_loss: 0.2173 - val_acc: 0.9188\n",
      "Epoch 22/50\n",
      "4232/4232 [==============================] - 3s 718us/step - loss: 0.1227 - acc: 0.9520 - val_loss: 0.1900 - val_acc: 0.9282\n",
      "Epoch 23/50\n",
      "4232/4232 [==============================] - 3s 677us/step - loss: 0.1188 - acc: 0.9537 - val_loss: 0.2247 - val_acc: 0.9254\n",
      "Epoch 24/50\n",
      "4232/4232 [==============================] - 3s 765us/step - loss: 0.1107 - acc: 0.9530 - val_loss: 0.2152 - val_acc: 0.9245\n",
      "Epoch 25/50\n",
      "4232/4232 [==============================] - 3s 761us/step - loss: 0.1009 - acc: 0.9631 - val_loss: 0.2087 - val_acc: 0.9348\n",
      "Epoch 26/50\n",
      "4232/4232 [==============================] - 4s 969us/step - loss: 0.1000 - acc: 0.9596 - val_loss: 0.1970 - val_acc: 0.9282\n",
      "Epoch 27/50\n",
      "4232/4232 [==============================] - 3s 820us/step - loss: 0.0944 - acc: 0.9608 - val_loss: 0.2189 - val_acc: 0.9207\n",
      "Epoch 28/50\n",
      "4232/4232 [==============================] - 4s 893us/step - loss: 0.0945 - acc: 0.9629 - val_loss: 0.1943 - val_acc: 0.9396\n",
      "Epoch 29/50\n",
      "4232/4232 [==============================] - 4s 895us/step - loss: 0.0899 - acc: 0.9672 - val_loss: 0.2005 - val_acc: 0.9282\n",
      "Epoch 30/50\n",
      "4232/4232 [==============================] - 3s 790us/step - loss: 0.0856 - acc: 0.9674 - val_loss: 0.2047 - val_acc: 0.9339\n",
      "Epoch 31/50\n",
      "4232/4232 [==============================] - 3s 758us/step - loss: 0.0794 - acc: 0.9683 - val_loss: 0.1887 - val_acc: 0.9367\n",
      "Epoch 32/50\n",
      "4232/4232 [==============================] - 3s 796us/step - loss: 0.0747 - acc: 0.9714 - val_loss: 0.2398 - val_acc: 0.9188\n",
      "Epoch 33/50\n",
      "4232/4232 [==============================] - 4s 853us/step - loss: 0.0862 - acc: 0.9657 - val_loss: 0.2088 - val_acc: 0.9377\n",
      "Epoch 34/50\n",
      "4232/4232 [==============================] - 4s 938us/step - loss: 0.0815 - acc: 0.9698 - val_loss: 0.2094 - val_acc: 0.9339\n",
      "Epoch 35/50\n",
      "4232/4232 [==============================] - 4s 870us/step - loss: 0.0662 - acc: 0.9752 - val_loss: 0.1766 - val_acc: 0.9386\n",
      "Epoch 36/50\n",
      "4232/4232 [==============================] - 4s 909us/step - loss: 0.0647 - acc: 0.9761 - val_loss: 0.2378 - val_acc: 0.9207\n",
      "Epoch 37/50\n",
      "4232/4232 [==============================] - 3s 797us/step - loss: 0.0795 - acc: 0.9672 - val_loss: 0.2214 - val_acc: 0.9282\n",
      "Epoch 38/50\n",
      "4232/4232 [==============================] - 3s 773us/step - loss: 0.0765 - acc: 0.9683 - val_loss: 0.2111 - val_acc: 0.9301\n",
      "Epoch 39/50\n",
      "4232/4232 [==============================] - 3s 826us/step - loss: 0.0666 - acc: 0.9738 - val_loss: 0.2099 - val_acc: 0.9377\n",
      "Epoch 40/50\n",
      "4232/4232 [==============================] - 3s 802us/step - loss: 0.0561 - acc: 0.9797 - val_loss: 0.1974 - val_acc: 0.9405\n",
      "Epoch 41/50\n",
      "4232/4232 [==============================] - 3s 740us/step - loss: 0.0605 - acc: 0.9754 - val_loss: 0.2198 - val_acc: 0.9235\n",
      "Epoch 42/50\n",
      "4232/4232 [==============================] - 3s 780us/step - loss: 0.0607 - acc: 0.9771 - val_loss: 0.2155 - val_acc: 0.9301\n",
      "Epoch 43/50\n",
      "4232/4232 [==============================] - 3s 796us/step - loss: 0.0531 - acc: 0.9797 - val_loss: 0.2069 - val_acc: 0.9367\n",
      "Epoch 44/50\n",
      "4232/4232 [==============================] - 3s 797us/step - loss: 0.0503 - acc: 0.9825 - val_loss: 0.1946 - val_acc: 0.9330\n",
      "Epoch 45/50\n",
      "4232/4232 [==============================] - 3s 763us/step - loss: 0.0660 - acc: 0.9698 - val_loss: 0.2619 - val_acc: 0.9178\n",
      "Epoch 46/50\n",
      "4232/4232 [==============================] - 3s 711us/step - loss: 0.0666 - acc: 0.9745 - val_loss: 0.2140 - val_acc: 0.9405\n",
      "Epoch 47/50\n",
      "4232/4232 [==============================] - 3s 699us/step - loss: 0.0584 - acc: 0.9773 - val_loss: 0.2059 - val_acc: 0.9415\n",
      "Epoch 48/50\n",
      "4232/4232 [==============================] - 3s 703us/step - loss: 0.0567 - acc: 0.9771 - val_loss: 0.2110 - val_acc: 0.9367ss: 0.0581 - acc:\n",
      "Epoch 49/50\n",
      "4232/4232 [==============================] - 3s 737us/step - loss: 0.0533 - acc: 0.9790 - val_loss: 0.2207 - val_acc: 0.9358\n",
      "Epoch 50/50\n",
      "4232/4232 [==============================] - 3s 724us/step - loss: 0.0553 - acc: 0.9802 - val_loss: 0.2008 - val_acc: 0.9424\n",
      "1059/1059 [==============================] - 0s 193us/step\n",
      "[INFO] loss=0.2008, accuracy: 94.2398%\n"
     ]
    }
   ],
   "source": [
    "model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])\n",
    "    \n",
    "# fit network\n",
    "history = model.fit(X_train, y_train, validation_data=(X_test, y_test), \n",
    "                    epochs=epochs, batch_size=batch_size, verbose=verbose)\n",
    "# evaluate model\n",
    "(loss, accuracy) = model.evaluate(X_test, y_test, batch_size=batch_size, verbose=verbose)\n",
    "print(\"[INFO] loss={:.4f}, accuracy: {:.4f}%\".format(loss, accuracy * 100))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.88      0.84      0.86        70\n",
      "           1       1.00      0.98      0.99       206\n",
      "           2       1.00      0.99      0.99       151\n",
      "           3       0.89      0.95      0.92       193\n",
      "           4       0.90      0.87      0.89       124\n",
      "           5       0.97      0.96      0.97       162\n",
      "           6       0.91      0.92      0.91       153\n",
      "\n",
      "   micro avg       0.94      0.94      0.94      1059\n",
      "   macro avg       0.94      0.93      0.93      1059\n",
      "weighted avg       0.94      0.94      0.94      1059\n",
      "\n"
     ]
    }
   ],
   "source": [
    "\n",
    "# Print confusion matrix for training data\n",
    "y_pred_train = model.predict(X_train)\n",
    "# Take the class with the highest probability from the train predictions\n",
    "max_y_pred_train = np.argmax(y_pred_train, axis=1)\n",
    "#print(classification_report(trainy, max_y_pred_train))\n",
    "\n",
    "# confusion matrix\n",
    "LABELS = [ 'Eating',\n",
    "            'Speak with Walk',\n",
    "          'Head Shake',\n",
    "          'Nodding',\n",
    "        'Speaking',      \n",
    "         'Stay', \n",
    "          'Walk']\n",
    "y_pred_test = model.predict(X_test)\n",
    "# Take the class with the highest probability from the test predictions\n",
    "max_y_pred_test = np.argmax(y_pred_test, axis=1)\n",
    "max_y_test = np.argmax(y_test, axis=1)\n",
    "\n",
    "matrix = metrics.confusion_matrix(max_y_test, max_y_pred_test)\n",
    "matrix = (matrix.astype('float') / matrix.sum(axis=1)[:, np.newaxis])*100\n",
    "fmt = '.2f'\n",
    "plt.figure(figsize=(8, 6))\n",
    "sns.heatmap(matrix,\n",
    "            cmap='RdYlGn_r',\n",
    "            linecolor='white',\n",
    "            linewidths=1,\n",
    "            xticklabels= LABELS,\n",
    "            yticklabels= LABELS,\n",
    "            annot=True,\n",
    "            fmt=fmt,\n",
    "           vmax = 100)\n",
    "plt.title('Confusion Matrix')\n",
    "plt.ylabel('True Label')\n",
    "plt.xlabel('Predicted Label')\n",
    "plt.show()\n",
    "\n",
    "print(classification_report(max_y_test, max_y_pred_test))"
   ]
  },
  {
   "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": "3.6.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}