LSTM_HAR.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/marcusnk237/lstm_har/blob/main/LSTM_HAR.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xK0iMxsg6ZmT"
      },
      "source": [
        "# <a title=\"Activity Recognition\" href=\"https://github.com/marcusnk237/lstm_har\" > LSTMs for Human Activity Recognition</a>\n",
        "\n",
        "Human Activity Recognition (HAR) using smartphones dataset and an LSTM RNN. Classifying the type of movement amongst six categories:\n",
        "- WALKING,\n",
        "- WALKING_UPSTAIRS,\n",
        "- WALKING_DOWNSTAIRS,\n",
        "- SITTING,\n",
        "- STANDING,\n",
        "- LAYING."
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Importing Libraries"
      ],
      "metadata": {
        "id": "KihEAiHXOt_m"
      }
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "kX3jIjAGiLHA",
        "outputId": "54f159fb-de77-4757-f081-9c68d9e99373"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Looking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/\n",
            "Collecting tensorflow_addons\n",
            "  Downloading tensorflow_addons-0.19.0-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (1.1 MB)\n",
            "\u001b[K     |████████████████████████████████| 1.1 MB 4.8 MB/s \n",
            "\u001b[?25hRequirement already satisfied: typeguard>=2.7 in /usr/local/lib/python3.8/dist-packages (from tensorflow_addons) (2.7.1)\n",
            "Requirement already satisfied: packaging in /usr/local/lib/python3.8/dist-packages (from tensorflow_addons) (21.3)\n",
            "Requirement already satisfied: pyparsing!=3.0.5,>=2.0.2 in /usr/local/lib/python3.8/dist-packages (from packaging->tensorflow_addons) (3.0.9)\n",
            "Installing collected packages: tensorflow-addons\n",
            "Successfully installed tensorflow-addons-0.19.0\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "WARNING:absl:`tf.distribute.experimental.TPUStrategy` is deprecated, please use  the non experimental symbol `tf.distribute.TPUStrategy` instead.\n"
          ]
        }
      ],
      "source": [
        "!pip install tensorflow_addons\n",
        "import cv2\n",
        "import gc\n",
        "import os\n",
        "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'\n",
        "from subprocess import call\n",
        "\n",
        "import tensorflow.compat.v1.keras.backend as K\n",
        "import tensorflow as tf\n",
        "\n",
        "from tensorflow.keras.callbacks import ModelCheckpoint, EarlyStopping\n",
        "from tensorflow import keras\n",
        "from tensorflow.keras.regularizers import l2\n",
        "from tensorflow_addons.optimizers import CyclicalLearningRate\n",
        "from tensorflow.keras import Model,Sequential,regularizers\n",
        "from tensorflow.keras.layers import MaxPool1D , Dense, Conv1D , Input , GaussianNoise , MaxPooling1D , LSTM, Flatten , add , Dropout , BatchNormalization , ReLU , GlobalAveragePooling1D\n",
        "\n",
        "import sklearn\n",
        "from sklearn import metrics\n",
        "from sklearn.model_selection import train_test_split, StratifiedKFold,StratifiedShuffleSplit\n",
        "from sklearn.metrics import classification_report\n",
        "from sklearn.utils import resample,shuffle\n",
        "from scipy import stats\n",
        "from sklearn.preprocessing import StandardScaler\n",
        "\n",
        "from collections import Counter\n",
        "from imblearn.over_sampling import SMOTE\n",
        "from random import choice\n",
        "import random \n",
        "# Generate and plot a synthetic imbalanced classification dataset\n",
        "import matplotlib.pyplot as plt\n",
        "import matplotlib as mpl\n",
        "mpl.style.use('seaborn')\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "import gc\n",
        "#### Import libraries\n",
        "\n",
        "import itertools\n",
        "import logging\n",
        "\n",
        "# detect and init the TPU \n",
        "tpu = tf.distribute.cluster_resolver.TPUClusterResolver.connect()\n",
        "# instantiate a distribution strategy\n",
        "tpu_strategy = tf.distribute.experimental.TPUStrategy(tpu)\n",
        "# Set random seed\n",
        "np.random.seed(123)"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Functions"
      ],
      "metadata": {
        "id": "iiKxkBJTQvEh"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Plotting confusion matrix:\n",
        "This function prints and plots the confusion matrix.\n",
        "Normalization can be applied by setting `normalize=True`."
      ],
      "metadata": {
        "id": "_FXa1srrQ1ba"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "def plot_confusion_matrix(cm, classes,\n",
        "                          normalize=False,\n",
        "                          title='Confusion matrix',\n",
        "                          cmap=plt.cm.Blues):\n",
        "    \"\"\"\n",
        "    This function prints and plots the confusion matrix.\n",
        "    Normalization can be applied by setting `normalize=True`.\n",
        "    \"\"\"\n",
        "    plt.imshow(cm, interpolation='nearest', cmap=cmap)\n",
        "    plt.title(title)\n",
        "    plt.colorbar()\n",
        "    tick_marks = np.arange(len(classes))\n",
        "    plt.xticks(tick_marks, classes, rotation=45)\n",
        "    plt.yticks(tick_marks, classes)\n",
        "\n",
        "    if normalize:\n",
        "        cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
        "        print(\"Normalized confusion matrix\")\n",
        "    else:\n",
        "        print('Confusion matrix, without normalization')\n",
        "\n",
        "    print(cm)\n",
        "\n",
        "    thresh = cm.max() / 2.\n",
        "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
        "        plt.text(j, i, cm[i, j],\n",
        "                 horizontalalignment=\"center\",\n",
        "                 color=\"white\" if cm[i, j] > thresh else \"black\")\n",
        "\n",
        "    plt.tight_layout()\n",
        "    plt.ylabel('True label')\n",
        "    plt.xlabel('Predicted label')\n",
        "    plt.savefig('/content/{}.jpg'.format(title), format='jpg', dpi=300)"
      ],
      "metadata": {
        "id": "jPJMK49tNBcU"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Plotting metrics curves\n",
        "This function prints and plots the training and test loss curve, as well as the training and test accuracy curve"
      ],
      "metadata": {
        "id": "1AtaFuJhQ_i6"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "def metrics_plot(history, field, fn):\n",
        "    def plot(data, val_data, best_index, best_value, title):\n",
        "        plt.plot(range(1, len(data)+1), data, label='train')\n",
        "        plt.plot(range(1, len(data)+1), val_data, label='validation')\n",
        "        if not best_index is None:\n",
        "            plt.axvline(x=best_index+1, linestyle=':', c=\"#777777\")\n",
        "        if not best_value is None:\n",
        "            plt.axhline(y=best_value, linestyle=':', c=\"#777777\")\n",
        "        plt.xlabel('Epoch')\n",
        "        plt.ylabel(field)\n",
        "        plt.xticks(range(0, len(data), 20))\n",
        "        plt.title(title)\n",
        "        plt.legend()\n",
        "        plt.savefig('/content/{}.jpg'.format(field), format='jpg', dpi=300)\n",
        "        plt.show()\n",
        "        \n",
        "        \n",
        "    try:\n",
        "       data = history.history[field]\n",
        "       val_data = history.history['val_' + field]\n",
        "    except:\n",
        "       data = history[field]\n",
        "       val_data = history['val_' + field]\n",
        "    tail = int(0.15 * len(data))\n",
        "\n",
        "    best_index = fn(val_data)\n",
        "    best_value = val_data[best_index]\n",
        "\n",
        "    plot(data, val_data, best_index, best_value, \"{} over epochs (best {:06.4f})\".format(field, best_value))\n",
        "    #plot(data[-tail:], val_data[-tail:], None, best_value, \"{} over last {} epochs\".format(field, tail))"
      ],
      "metadata": {
        "id": "huh8wqdRNJPI"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Human activity recognition model"
      ],
      "metadata": {
        "id": "FzeF6hKURMSs"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "def get_har_model(timesteps, input_dim):    \n",
        "    inputs_cnn=keras.layers.Input(shape=(timesteps, input_dim), name='inputs_cnn')\n",
        "    pool4 = keras.layers.LSTM(128,return_sequences=True)(inputs_cnn)\n",
        "    pool4=  keras.layers.Dropout(0.5)(pool4)\n",
        "    pool4 = keras.layers.LSTM(128,return_sequences=True)(pool4)\n",
        "    pool4=  keras.layers.Dropout(0.5)(pool4)\n",
        "    pool4=keras.layers.Flatten()(pool4)\n",
        "    pool4 = keras.layers.Dense(64, activation='relu')(pool4)\n",
        "    main_output = keras.layers.Dense(n_classes, activation='softmax', name='main_output')(pool4)\n",
        "    model = keras.Model(inputs= inputs_cnn, outputs=main_output)\n",
        "    return model"
      ],
      "metadata": {
        "id": "N-rtCppFA7ex"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VPKjP-mn6ZmZ"
      },
      "source": [
        "## Preparing dataset"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Signals labels"
      ],
      "metadata": {
        "id": "Lj3Kz5W3OzRD"
      }
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "n_fDXYnz6ZmX"
      },
      "outputs": [],
      "source": [
        "# Useful Constants\n",
        "\n",
        "# Those are separate normalised input features for the neural network\n",
        "INPUT_SIGNAL_TYPES = [\n",
        "    \"body_acc_x_\",\n",
        "    \"body_acc_y_\",\n",
        "    \"body_acc_z_\",\n",
        "    \"body_gyro_x_\",\n",
        "    \"body_gyro_y_\",\n",
        "    \"body_gyro_z_\",\n",
        "    \"total_acc_x_\",\n",
        "    \"total_acc_y_\",\n",
        "    \"total_acc_z_\"\n",
        "]\n",
        "\n",
        "# Output classes to learn how to classify\n",
        "LABELS = [\n",
        "    \"WALKING\", \n",
        "    \"WALKING_UPSTAIRS\", \n",
        "    \"WALKING_DOWNSTAIRS\", \n",
        "    \"SITTING\", \n",
        "    \"STANDING\", \n",
        "    \"LAYING\"\n",
        "] \n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "SIntKTuj6ZmX"
      },
      "source": [
        "### Downloading datas"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(\"\")\n",
        "\n",
        "print(\"Downloading...\")\n",
        "if not os.path.exists(\"UCI HAR Dataset.zip\"):\n",
        "    call(\n",
        "        'wget \"https://archive.ics.uci.edu/ml/machine-learning-databases/00240/UCI HAR Dataset.zip\"',\n",
        "        shell=True\n",
        "    )\n",
        "    print(\"Downloading done.\\n\")\n",
        "else:\n",
        "    print(\"Dataset already downloaded. Did not download twice.\\n\")\n",
        "\n",
        "\n",
        "print(\"Extracting...\")\n",
        "extract_directory = os.path.abspath(\"UCI HAR Dataset\")\n",
        "if not os.path.exists(extract_directory):\n",
        "    call(\n",
        "        'unzip -nq \"UCI HAR Dataset.zip\"',\n",
        "        shell=True\n",
        "    )\n",
        "    print(\"Extracting successfully done to {}.\".format(extract_directory))\n",
        "else:\n",
        "    print(\"Dataset already extracted. Did not extract twice.\\n\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "JvnG6npV7K69",
        "outputId": "378f450b-5740-499d-a3fb-84912c4b9f32"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\n",
            "Downloading...\n",
            "Downloading done.\n",
            "\n",
            "Extracting...\n",
            "Extracting successfully done to /content/UCI HAR Dataset.\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Loading datas"
      ],
      "metadata": {
        "id": "BRXoN_3MQIXZ"
      }
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "o0EO_xda6ZmZ"
      },
      "outputs": [],
      "source": [
        "TRAIN = \"train/\"\n",
        "TEST = \"test/\"\n",
        "\n",
        "DATASET_PATH='/content/UCI HAR Dataset/'\n",
        "# Load \"X\" (the neural network's training and testing inputs)\n",
        "\n",
        "def load_X(X_signals_paths):\n",
        "    X_signals = []\n",
        "    \n",
        "    for signal_type_path in X_signals_paths:\n",
        "        file = open(signal_type_path, 'r')\n",
        "        # Read dataset from disk, dealing with text files' syntax\n",
        "        X_signals.append(\n",
        "            [np.array(serie, dtype=np.float32) for serie in [\n",
        "                row.replace('  ', ' ').strip().split(' ') for row in file\n",
        "            ]]\n",
        "        )\n",
        "        file.close()\n",
        "    \n",
        "    return np.transpose(np.array(X_signals), (1, 2, 0))\n",
        "\n",
        "X_train_signals_paths = [\n",
        "    DATASET_PATH + TRAIN + \"Inertial Signals/\" + signal + \"train.txt\" for signal in INPUT_SIGNAL_TYPES\n",
        "]\n",
        "X_test_signals_paths = [\n",
        "    DATASET_PATH + TEST + \"Inertial Signals/\" + signal + \"test.txt\" for signal in INPUT_SIGNAL_TYPES\n",
        "]\n",
        "\n",
        "X_train = load_X(X_train_signals_paths)\n",
        "X_test = load_X(X_test_signals_paths)\n",
        "\n",
        "\n",
        "# Load \"y\" (the neural network's training and testing outputs)\n",
        "\n",
        "def load_y(y_path):\n",
        "    file = open(y_path, 'r')\n",
        "    # Read dataset from disk, dealing with text file's syntax\n",
        "    y_ = np.array(\n",
        "        [elem for elem in [\n",
        "            row.replace('  ', ' ').strip().split(' ') for row in file\n",
        "        ]], \n",
        "        dtype=np.int32\n",
        "    )\n",
        "    file.close()\n",
        "    \n",
        "    # Substract 1 to each output class for friendly 0-based indexing \n",
        "    return y_ - 1\n",
        "\n",
        "y_train_path = DATASET_PATH + TRAIN + \"y_train.txt\"\n",
        "y_test_path = DATASET_PATH + TEST + \"y_test.txt\"\n",
        "\n",
        "y_train = load_y(y_train_path)\n",
        "y_test = load_y(y_test_path)\n"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Training and test datas\n",
        "Since to training isn't sufficient enough (Around 70% of the global dataset), we processed as follow:\n",
        "- We reassemble the dataset\n",
        "- Then, we split it into: Training dataset, test dataset and validation dataset"
      ],
      "metadata": {
        "id": "ZhcDqjv5FqOp"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "##### Reassembling data"
      ],
      "metadata": {
        "id": "HuPk6_KNPmYj"
      }
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "ktmH3YNT6Zma"
      },
      "outputs": [],
      "source": [
        "# Input Data \n",
        "\n",
        "training_data_count = len(X_train)  # 7352 training series (with 50% overlap between each serie)\n",
        "test_data_count = len(X_test)  # 2947 testing series\n",
        "n_steps = len(X_train[0])  # 128 timesteps per series\n",
        "n_input = len(X_train[0][0])  # 9 input parameters per timestep\n",
        "n_classes = len(LABELS)"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "X=np.concatenate((X_train,X_test),axis=0) ### Reassembling datas ( features)\n",
        "Y=np.concatenate((y_train,y_test), axis=0) ### Reassembling datas ( labels)\n",
        "Y= np.array([i[0] for i in Y])"
      ],
      "metadata": {
        "id": "7KCZMRXAEldJ"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "##### Split into train, test and validation"
      ],
      "metadata": {
        "id": "qXxRh7DgPpbi"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "Y=Y.astype(int) # Convert label into int\n",
        "## Split into train and test\n",
        "X_train, X_val, y_train, y_val = train_test_split(X,Y, test_size=0.1, random_state=np.random.randint(1,1000, 1)[0], shuffle = True)\n",
        "## Slip into test and validation\n",
        "X_val, X_test, y_val, y_test  = train_test_split(X_val, y_val, test_size=0.5, random_state=np.random.randint(1,1000, 1)[0], shuffle = True)"
      ],
      "metadata": {
        "id": "FF8viaaXFp4u"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Training and Testing"
      ],
      "metadata": {
        "id": "7Hj5XX0eQcth"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Training"
      ],
      "metadata": {
        "id": "nrfoIajBQm4r"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "name_weights = \"human_activity_recognition_weights.h5\"\n",
        "N=X_train.shape[0]\n",
        "batch_size = 32 * tpu_strategy.num_replicas_in_sync\n",
        "iterations = N/batch_size\n",
        "step_size= 2 * iterations\n",
        "lr_schedule = CyclicalLearningRate(1e-6, 1e-3, step_size=step_size, scale_fn=lambda x: tf.pow(0.95,x))\n",
        "optimizer = tf.keras.optimizers.Adam(learning_rate=lr_schedule)\n",
        "print(batch_size) \n",
        "with tpu_strategy.scope():\n",
        "        har_model = get_har_model(X_train.shape[1],X_train.shape[2]) # Instanciate ResVidNet\n",
        "        early_stopping = EarlyStopping(monitor='val_loss', patience=5, verbose=2) # Early stopping parameters\n",
        "        save_best_weights = ModelCheckpoint(filepath=name_weights, verbose=2, save_best_only=True , restore_best_weights=True) # Model checkpoint parameters\n",
        "        har_model.compile(optimizer=optimizer, loss='sparse_categorical_crossentropy', metrics=['accuracy']) # Model compiling\n",
        "history = har_model.fit(X_train, y_train, validation_data=(X_test, y_test), batch_size=batch_size, epochs=100, shuffle=False,callbacks=[save_best_weights,early_stopping])\n",
        "np.save(\"/content/har_history_.npy\",history.history)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "H4RHBTZbAsyw",
        "outputId": "48e59384-dfc0-47fa-f6bf-c2f02a240293"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "256\n",
            "Epoch 1/100\n",
            "37/37 [==============================] - ETA: 0s - loss: 1.3042 - accuracy: 0.4619\n",
            "Epoch 1: val_loss improved from inf to 0.89562, saving model to /content/drive/MyDrive/RESVIDLSTM/Models/HAR/human_activity_recognition_weights.h5\n",
            "37/37 [==============================] - 15s 207ms/step - loss: 1.3042 - accuracy: 0.4619 - val_loss: 0.8956 - val_accuracy: 0.5903\n",
            "Epoch 2/100\n",
            "37/37 [==============================] - ETA: 0s - loss: 0.7062 - accuracy: 0.7060\n",
            "Epoch 2: val_loss improved from 0.89562 to 0.40274, saving model to /content/drive/MyDrive/RESVIDLSTM/Models/HAR/human_activity_recognition_weights.h5\n",
            "37/37 [==============================] - 2s 55ms/step - loss: 0.7062 - accuracy: 0.7060 - val_loss: 0.4027 - val_accuracy: 0.8447\n",
            "Epoch 3/100\n",
            "37/37 [==============================] - ETA: 0s - loss: 0.3360 - accuracy: 0.8775\n",
            "Epoch 3: val_loss improved from 0.40274 to 0.23149, saving model to /content/drive/MyDrive/RESVIDLSTM/Models/HAR/human_activity_recognition_weights.h5\n",
            "37/37 [==============================] - 2s 55ms/step - loss: 0.3360 - accuracy: 0.8775 - val_loss: 0.2315 - val_accuracy: 0.9049\n",
            "Epoch 4/100\n",
            "37/37 [==============================] - ETA: 0s - loss: 0.2388 - accuracy: 0.9113\n",
            "Epoch 4: val_loss improved from 0.23149 to 0.17465, saving model to /content/drive/MyDrive/RESVIDLSTM/Models/HAR/human_activity_recognition_weights.h5\n",
            "37/37 [==============================] - 2s 55ms/step - loss: 0.2388 - accuracy: 0.9113 - val_loss: 0.1746 - val_accuracy: 0.9359\n",
            "Epoch 5/100\n",
            "37/37 [==============================] - ETA: 0s - loss: 0.2095 - accuracy: 0.9224\n",
            "Epoch 5: val_loss improved from 0.17465 to 0.15928, saving model to /content/drive/MyDrive/RESVIDLSTM/Models/HAR/human_activity_recognition_weights.h5\n",
            "37/37 [==============================] - 2s 55ms/step - loss: 0.2095 - accuracy: 0.9224 - val_loss: 0.1593 - val_accuracy: 0.9301\n",
            "Epoch 6/100\n",
            "37/37 [==============================] - ETA: 0s - loss: 0.1668 - accuracy: 0.9325\n",
            "Epoch 6: val_loss improved from 0.15928 to 0.14361, saving model to /content/drive/MyDrive/RESVIDLSTM/Models/HAR/human_activity_recognition_weights.h5\n",
            "37/37 [==============================] - 2s 55ms/step - loss: 0.1668 - accuracy: 0.9325 - val_loss: 0.1436 - val_accuracy: 0.9437\n",
            "Epoch 7/100\n",
            "37/37 [==============================] - ETA: 0s - loss: 0.1427 - accuracy: 0.9414\n",
            "Epoch 7: val_loss improved from 0.14361 to 0.11507, saving model to /content/drive/MyDrive/RESVIDLSTM/Models/HAR/human_activity_recognition_weights.h5\n",
            "37/37 [==============================] - 2s 56ms/step - loss: 0.1427 - accuracy: 0.9414 - val_loss: 0.1151 - val_accuracy: 0.9534\n",
            "Epoch 8/100\n",
            "37/37 [==============================] - ETA: 0s - loss: 0.1177 - accuracy: 0.9508\n",
            "Epoch 8: val_loss improved from 0.11507 to 0.09744, saving model to /content/drive/MyDrive/RESVIDLSTM/Models/HAR/human_activity_recognition_weights.h5\n",
            "37/37 [==============================] - 3s 82ms/step - loss: 0.1177 - accuracy: 0.9508 - val_loss: 0.0974 - val_accuracy: 0.9553\n",
            "Epoch 9/100\n",
            "37/37 [==============================] - ETA: 0s - loss: 0.1148 - accuracy: 0.9521\n",
            "Epoch 9: val_loss did not improve from 0.09744\n",
            "37/37 [==============================] - 2s 50ms/step - loss: 0.1148 - accuracy: 0.9521 - val_loss: 0.1070 - val_accuracy: 0.9515\n",
            "Epoch 10/100\n",
            "37/37 [==============================] - ETA: 0s - loss: 0.1284 - accuracy: 0.9480\n",
            "Epoch 10: val_loss did not improve from 0.09744\n",
            "37/37 [==============================] - 2s 49ms/step - loss: 0.1284 - accuracy: 0.9480 - val_loss: 0.1262 - val_accuracy: 0.9495\n",
            "Epoch 11/100\n",
            "37/37 [==============================] - ETA: 0s - loss: 0.1262 - accuracy: 0.9488\n",
            "Epoch 11: val_loss did not improve from 0.09744\n",
            "37/37 [==============================] - 2s 49ms/step - loss: 0.1262 - accuracy: 0.9488 - val_loss: 0.1154 - val_accuracy: 0.9495\n",
            "Epoch 12/100\n",
            "37/37 [==============================] - ETA: 0s - loss: 0.1172 - accuracy: 0.9530\n",
            "Epoch 12: val_loss improved from 0.09744 to 0.09584, saving model to /content/drive/MyDrive/RESVIDLSTM/Models/HAR/human_activity_recognition_weights.h5\n",
            "37/37 [==============================] - 2s 55ms/step - loss: 0.1172 - accuracy: 0.9530 - val_loss: 0.0958 - val_accuracy: 0.9631\n",
            "Epoch 13/100\n",
            "37/37 [==============================] - ETA: 0s - loss: 0.1140 - accuracy: 0.9517\n",
            "Epoch 13: val_loss did not improve from 0.09584\n",
            "37/37 [==============================] - 2s 50ms/step - loss: 0.1140 - accuracy: 0.9517 - val_loss: 0.1308 - val_accuracy: 0.9456\n",
            "Epoch 14/100\n",
            "37/37 [==============================] - ETA: 0s - loss: 0.1093 - accuracy: 0.9541\n",
            "Epoch 14: val_loss did not improve from 0.09584\n",
            "37/37 [==============================] - 2s 50ms/step - loss: 0.1093 - accuracy: 0.9541 - val_loss: 0.1170 - val_accuracy: 0.9515\n",
            "Epoch 15/100\n",
            "37/37 [==============================] - ETA: 0s - loss: 0.1085 - accuracy: 0.9551\n",
            "Epoch 15: val_loss did not improve from 0.09584\n",
            "37/37 [==============================] - 2s 49ms/step - loss: 0.1085 - accuracy: 0.9551 - val_loss: 0.1013 - val_accuracy: 0.9573\n",
            "Epoch 16/100\n",
            "36/37 [============================>.] - ETA: 0s - loss: 0.1034 - accuracy: 0.9545\n",
            "Epoch 16: val_loss did not improve from 0.09584\n",
            "37/37 [==============================] - 2s 52ms/step - loss: 0.1030 - accuracy: 0.9547 - val_loss: 0.0971 - val_accuracy: 0.9573\n",
            "Epoch 17/100\n",
            "37/37 [==============================] - ETA: 0s - loss: 0.1028 - accuracy: 0.9562\n",
            "Epoch 17: val_loss did not improve from 0.09584\n",
            "37/37 [==============================] - 2s 48ms/step - loss: 0.1028 - accuracy: 0.9562 - val_loss: 0.1089 - val_accuracy: 0.9553\n",
            "Epoch 17: early stopping\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Testing"
      ],
      "metadata": {
        "id": "5lyOuqPCRVMl"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "har_model.load_weights(\"name_weights) \n",
        "X_covid_test = har_model.evaluate(X_val, y_val) # Evaluate model"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "2HReVVhINThR",
        "outputId": "8821be42-ec90-40cb-a2bb-ac44806356cb"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "17/17 [==============================] - 1s 21ms/step - loss: 0.1142 - accuracy: 0.9573\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Plotting metrics"
      ],
      "metadata": {
        "id": "bj1BjtLTRWy0"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "metrics_plot(history, 'loss', lambda x: np.argmin(x))\n",
        "metrics_plot(history, 'accuracy', lambda x: np.argmax(x))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 701
        },
        "id": "v3AwRRCDNqW-",
        "outputId": "d3c2b461-9707-4320-c287-e8d7e9a546ad"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Confusion matrix"
      ],
      "metadata": {
        "id": "WGZppNyURZf4"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "y_pred = tf.argmax(har_model.predict(X_val), axis=-1)\n",
        "# Compute confusion matrix\n",
        "cnf_matrix = sklearn.metrics.confusion_matrix(y_val, y_pred)\n",
        "np.set_printoptions(precision=2)\n",
        "\n",
        "\n",
        "# Plot non-normalized confusion matrix\n",
        "plt.figure()\n",
        "plot_confusion_matrix(cnf_matrix, classes=LABELS,title=' Confusion matrix')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 558
        },
        "id": "cJTMUyBaNr-9",
        "outputId": "35e47b31-e31f-4124-f51f-09bfcae3a8a8"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "17/17 [==============================] - 3s 92ms/step\n",
            "Confusion matrix, without normalization\n",
            "[[ 90   0   0   0   1   0]\n",
            " [  0  77   0   0   0   0]\n",
            " [  0   0  60   0   0   0]\n",
            " [  0   0   0  87   9   0]\n",
            " [  0   0   0  12  68   0]\n",
            " [  0   0   0   0   0 111]]\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 576x396 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Classification report"
      ],
      "metadata": {
        "id": "fj2Qb95uRcvN"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "df_class_r = pd.DataFrame(classification_report(y_pred, y_val, digits=2,output_dict=True)).T\n",
        "df_class_r['support'] = df_class_r.support.apply(int)\n",
        "df_class_r.rename(index={'0':LABELS[0], '1':LABELS[1], '2':LABELS[2] ,'3':LABELS[3],'4': LABELS[4],'5': LABELS[5]}, inplace=True)\n",
        "df_r= df_class_r\n",
        "df_r.style.background_gradient(cmap='viridis')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 332
        },
        "id": "qmvJ-djJN24H",
        "outputId": "77d5257f-869e-4dcc-d814-b62c13ad8569"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<pandas.io.formats.style.Styler at 0x7f161d15d970>"
            ],
            "text/html": [
              "<style type=\"text/css\">\n",
              "#T_04258_row0_col0 {\n",
              "  background-color: #d0e11c;\n",
              "  color: #000000;\n",
              "}\n",
              "#T_04258_row0_col1, #T_04258_row1_col0, #T_04258_row1_col1, #T_04258_row1_col2, #T_04258_row2_col0, #T_04258_row2_col1, #T_04258_row2_col2, #T_04258_row5_col0, #T_04258_row5_col1, #T_04258_row5_col2, #T_04258_row7_col3, #T_04258_row8_col3 {\n",
              "  background-color: #fde725;\n",
              "  color: #000000;\n",
              "}\n",
              "#T_04258_row0_col2 {\n",
              "  background-color: #e5e419;\n",
              "  color: #000000;\n",
              "}\n",
              "#T_04258_row0_col3 {\n",
              "  background-color: #443b84;\n",
              "  color: #f1f1f1;\n",
              "}\n",
              "#T_04258_row1_col3, #T_04258_row4_col3 {\n",
              "  background-color: #463480;\n",
              "  color: #f1f1f1;\n",
              "}\n",
              "#T_04258_row2_col3 {\n",
              "  background-color: #482979;\n",
              "  color: #f1f1f1;\n",
              "}\n",
              "#T_04258_row3_col0 {\n",
              "  background-color: #2c728e;\n",
              "  color: #f1f1f1;\n",
              "}\n",
              "#T_04258_row3_col1 {\n",
              "  background-color: #481467;\n",
              "  color: #f1f1f1;\n",
              "}\n",
              "#T_04258_row3_col2 {\n",
              "  background-color: #3e4c8a;\n",
              "  color: #f1f1f1;\n",
              "}\n",
              "#T_04258_row3_col3 {\n",
              "  background-color: #424186;\n",
              "  color: #f1f1f1;\n",
              "}\n",
              "#T_04258_row4_col0, #T_04258_row4_col1, #T_04258_row4_col2, #T_04258_row6_col3 {\n",
              "  background-color: #440154;\n",
              "  color: #f1f1f1;\n",
              "}\n",
              "#T_04258_row5_col3 {\n",
              "  background-color: #3f4889;\n",
              "  color: #f1f1f1;\n",
              "}\n",
              "#T_04258_row6_col0, #T_04258_row7_col0, #T_04258_row8_col0 {\n",
              "  background-color: #4cc26c;\n",
              "  color: #000000;\n",
              "}\n",
              "#T_04258_row6_col1, #T_04258_row8_col1 {\n",
              "  background-color: #35b779;\n",
              "  color: #f1f1f1;\n",
              "}\n",
              "#T_04258_row6_col2, #T_04258_row8_col2 {\n",
              "  background-color: #40bd72;\n",
              "  color: #f1f1f1;\n",
              "}\n",
              "#T_04258_row7_col1 {\n",
              "  background-color: #38b977;\n",
              "  color: #f1f1f1;\n",
              "}\n",
              "#T_04258_row7_col2 {\n",
              "  background-color: #42be71;\n",
              "  color: #f1f1f1;\n",
              "}\n",
              "</style>\n",
              "<table id=\"T_04258_\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr>\n",
              "      <th class=\"blank level0\" >&nbsp;</th>\n",
              "      <th class=\"col_heading level0 col0\" >precision</th>\n",
              "      <th class=\"col_heading level0 col1\" >recall</th>\n",
              "      <th class=\"col_heading level0 col2\" >f1-score</th>\n",
              "      <th class=\"col_heading level0 col3\" >support</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th id=\"T_04258_level0_row0\" class=\"row_heading level0 row0\" >WALKING</th>\n",
              "      <td id=\"T_04258_row0_col0\" class=\"data row0 col0\" >0.989011</td>\n",
              "      <td id=\"T_04258_row0_col1\" class=\"data row0 col1\" >1.000000</td>\n",
              "      <td id=\"T_04258_row0_col2\" class=\"data row0 col2\" >0.994475</td>\n",
              "      <td id=\"T_04258_row0_col3\" class=\"data row0 col3\" >90</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th id=\"T_04258_level0_row1\" class=\"row_heading level0 row1\" >WALKING_UPSTAIRS</th>\n",
              "      <td id=\"T_04258_row1_col0\" class=\"data row1 col0\" >1.000000</td>\n",
              "      <td id=\"T_04258_row1_col1\" class=\"data row1 col1\" >1.000000</td>\n",
              "      <td id=\"T_04258_row1_col2\" class=\"data row1 col2\" >1.000000</td>\n",
              "      <td id=\"T_04258_row1_col3\" class=\"data row1 col3\" >77</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th id=\"T_04258_level0_row2\" class=\"row_heading level0 row2\" >WALKING_DOWNSTAIRS</th>\n",
              "      <td id=\"T_04258_row2_col0\" class=\"data row2 col0\" >1.000000</td>\n",
              "      <td id=\"T_04258_row2_col1\" class=\"data row2 col1\" >1.000000</td>\n",
              "      <td id=\"T_04258_row2_col2\" class=\"data row2 col2\" >1.000000</td>\n",
              "      <td id=\"T_04258_row2_col3\" class=\"data row2 col3\" >60</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th id=\"T_04258_level0_row3\" class=\"row_heading level0 row3\" >SITTING</th>\n",
              "      <td id=\"T_04258_row3_col0\" class=\"data row3 col0\" >0.906250</td>\n",
              "      <td id=\"T_04258_row3_col1\" class=\"data row3 col1\" >0.878788</td>\n",
              "      <td id=\"T_04258_row3_col2\" class=\"data row3 col2\" >0.892308</td>\n",
              "      <td id=\"T_04258_row3_col3\" class=\"data row3 col3\" >99</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th id=\"T_04258_level0_row4\" class=\"row_heading level0 row4\" >STANDING</th>\n",
              "      <td id=\"T_04258_row4_col0\" class=\"data row4 col0\" >0.850000</td>\n",
              "      <td id=\"T_04258_row4_col1\" class=\"data row4 col1\" >0.871795</td>\n",
              "      <td id=\"T_04258_row4_col2\" class=\"data row4 col2\" >0.860759</td>\n",
              "      <td id=\"T_04258_row4_col3\" class=\"data row4 col3\" >78</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th id=\"T_04258_level0_row5\" class=\"row_heading level0 row5\" >LAYING</th>\n",
              "      <td id=\"T_04258_row5_col0\" class=\"data row5 col0\" >1.000000</td>\n",
              "      <td id=\"T_04258_row5_col1\" class=\"data row5 col1\" >1.000000</td>\n",
              "      <td id=\"T_04258_row5_col2\" class=\"data row5 col2\" >1.000000</td>\n",
              "      <td id=\"T_04258_row5_col3\" class=\"data row5 col3\" >111</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th id=\"T_04258_level0_row6\" class=\"row_heading level0 row6\" >accuracy</th>\n",
              "      <td id=\"T_04258_row6_col0\" class=\"data row6 col0\" >0.957282</td>\n",
              "      <td id=\"T_04258_row6_col1\" class=\"data row6 col1\" >0.957282</td>\n",
              "      <td id=\"T_04258_row6_col2\" class=\"data row6 col2\" >0.957282</td>\n",
              "      <td id=\"T_04258_row6_col3\" class=\"data row6 col3\" >0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th id=\"T_04258_level0_row7\" class=\"row_heading level0 row7\" >macro avg</th>\n",
              "      <td id=\"T_04258_row7_col0\" class=\"data row7 col0\" >0.957543</td>\n",
              "      <td id=\"T_04258_row7_col1\" class=\"data row7 col1\" >0.958430</td>\n",
              "      <td id=\"T_04258_row7_col2\" class=\"data row7 col2\" >0.957924</td>\n",
              "      <td id=\"T_04258_row7_col3\" class=\"data row7 col3\" >515</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th id=\"T_04258_level0_row8\" class=\"row_heading level0 row8\" >weighted avg</th>\n",
              "      <td id=\"T_04258_row8_col0\" class=\"data row8 col0\" >0.957339</td>\n",
              "      <td id=\"T_04258_row8_col1\" class=\"data row8 col1\" >0.957282</td>\n",
              "      <td id=\"T_04258_row8_col2\" class=\"data row8 col2\" >0.957244</td>\n",
              "      <td id=\"T_04258_row8_col3\" class=\"data row8 col3\" >515</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n"
            ]
          },
          "metadata": {},
          "execution_count": 96
        }
      ]
    }
  ],
  "metadata": {
    "hide_input": false,
    "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": "3.6.8"
    },
    "colab": {
      "provenance": [],
      "toc_visible": true,
      "machine_shape": "hm",
      "include_colab_link": true
    },
    "accelerator": "TPU",
    "gpuClass": "standard"
  },
  "nbformat": 4,
  "nbformat_minor": 0
}