test_CNN_model.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Human Activity Recognition Accelerometer.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/AMEYABHAVE/Cristl/blob/main/test_CNN_model.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "BhdmLvLFcmVR",
        "outputId": "c12195a8-f9bc-4db9-fce6-4ab4d362544a"
      },
      "source": [
        "import tensorflow as tf\n",
        "from tensorflow.keras import Sequential\n",
        "from tensorflow.keras.layers import Flatten, Dense, Dropout, BatchNormalization\n",
        "from tensorflow.keras.layers import Conv2D, MaxPool2D\n",
        "from tensorflow.keras.optimizers import Adam\n",
        "print(tf.__version__)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "2.6.0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Tlteb_NUc103"
      },
      "source": [
        "from google.colab import drive\n",
        "import pandas as pd\n",
        "file_location = '/content/drive/MyDrive/accelerometer.csv'\n",
        "accelerometer =pd.read_csv(file_location)\n",
        "accelerometer.drop(columns=['time','repetition','uuid'], inplace=True)\n",
        "accchair= accelerometer[accelerometer['activity'] == 'CHAIR'].head(22743)\n",
        "acccouch= accelerometer[accelerometer['activity'] == 'COUCH'].head(22743)\n",
        "acclyingside= accelerometer[accelerometer['activity'] == 'LYING_SIDE'].head(22743)\n",
        "accwalk= accelerometer[accelerometer['activity'] == 'WALKING'].head(22743)\n",
        "accdown= accelerometer[accelerometer['activity'] == 'DOWNSTAIRS'].head(22743)\n",
        "accup= accelerometer[accelerometer['activity'] == 'UPSTAIRS'].head(22743)\n",
        "accrunning= accelerometer[accelerometer['activity'] == 'RUNNING']\n",
        "accstanding= accelerometer[accelerometer['activity'] == 'STANDING'].head(22743)\n",
        "acclyingup= accelerometer[accelerometer['activity'] == 'LYING_UP'].head(22743)\n",
        "\n",
        "accdataset = pd.DataFrame()\n",
        "accdataset = accdataset.append([accchair, acclyingside, acclyingup, accrunning])\n",
        "#accdataset = accdataset.append([accchair, acccouch, acclyingside, accwalk, accdown, accup, accstanding, acclyingup, accrunning])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "OAi32HfrfWwb",
        "outputId": "2a576002-0684-428d-f286-f721af490caa"
      },
      "source": [
        "accdataset['activity'].value_counts()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "LYING_SIDE    22743\n",
              "RUNNING       22743\n",
              "LYING_UP      22743\n",
              "CHAIR         22743\n",
              "Name: activity, dtype: int64"
            ]
          },
          "metadata": {},
          "execution_count": 3
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "VjHHlnRhf2ER",
        "outputId": "1a20810e-bf03-4879-844b-b1f5a3feb0b4"
      },
      "source": [
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.preprocessing import StandardScaler, LabelEncoder\n",
        "alabel=LabelEncoder()\n",
        "accdataset['label']=alabel.fit_transform(accdataset['activity'])\n",
        "print(accdataset.head())\n",
        "print(alabel.classes_)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "            name     acc_x     acc_y     acc_z activity  label\n",
            "0  accelerometer -0.414825  7.590469  6.235336    CHAIR      0\n",
            "1  accelerometer -0.337662  7.645508  6.125275    CHAIR      0\n",
            "2  accelerometer -0.393295  7.605423  6.231155    CHAIR      0\n",
            "3  accelerometer -0.469269  7.592865  6.281403    CHAIR      0\n",
            "4  accelerometer -0.418411  7.633530  6.317307    CHAIR      0\n",
            "['CHAIR' 'LYING_SIDE' 'LYING_UP' 'RUNNING']\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 417
        },
        "id": "S-fji2OCf9d4",
        "outputId": "b6ffc090-7c97-449a-a756-3e81fdaab899"
      },
      "source": [
        "AX = accdataset[['acc_x','acc_y','acc_z']]\n",
        "AY = accdataset['label']\n",
        "scaler = StandardScaler()\n",
        "AX = scaler.fit_transform(AX)\n",
        "Ascaled_X = pd.DataFrame(data=AX, columns=['acc_x','acc_y','acc_z'])\n",
        "Ascaled_X['label'] = AY.values\n",
        "Ascaled_X"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "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>acc_x</th>\n",
              "      <th>acc_y</th>\n",
              "      <th>acc_z</th>\n",
              "      <th>label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>-0.265185</td>\n",
              "      <td>0.574743</td>\n",
              "      <td>0.555825</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>-0.246453</td>\n",
              "      <td>0.592224</td>\n",
              "      <td>0.536195</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>-0.259959</td>\n",
              "      <td>0.579492</td>\n",
              "      <td>0.555080</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>-0.278402</td>\n",
              "      <td>0.575504</td>\n",
              "      <td>0.564042</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>-0.266056</td>\n",
              "      <td>0.588420</td>\n",
              "      <td>0.570445</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>90967</th>\n",
              "      <td>-0.329373</td>\n",
              "      <td>-0.767396</td>\n",
              "      <td>1.128777</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>90968</th>\n",
              "      <td>-0.445554</td>\n",
              "      <td>-0.970886</td>\n",
              "      <td>1.096663</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>90969</th>\n",
              "      <td>-0.386593</td>\n",
              "      <td>-1.146441</td>\n",
              "      <td>0.662523</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>90970</th>\n",
              "      <td>-0.366553</td>\n",
              "      <td>-1.287800</td>\n",
              "      <td>0.643211</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>90971</th>\n",
              "      <td>-0.717989</td>\n",
              "      <td>-1.335488</td>\n",
              "      <td>0.912081</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>90972 rows × 4 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "          acc_x     acc_y     acc_z  label\n",
              "0     -0.265185  0.574743  0.555825      0\n",
              "1     -0.246453  0.592224  0.536195      0\n",
              "2     -0.259959  0.579492  0.555080      0\n",
              "3     -0.278402  0.575504  0.564042      0\n",
              "4     -0.266056  0.588420  0.570445      0\n",
              "...         ...       ...       ...    ...\n",
              "90967 -0.329373 -0.767396  1.128777      3\n",
              "90968 -0.445554 -0.970886  1.096663      3\n",
              "90969 -0.386593 -1.146441  0.662523      3\n",
              "90970 -0.366553 -1.287800  0.643211      3\n",
              "90971 -0.717989 -1.335488  0.912081      3\n",
              "\n",
              "[90972 rows x 4 columns]"
            ]
          },
          "metadata": {},
          "execution_count": 5
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "f_lnSj_ZgA9V"
      },
      "source": [
        "import scipy.stats as stats\n",
        "Fs = 20\n",
        "frame_s = Fs*2\n",
        "hop_s = Fs"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DthwrqQcgDPS"
      },
      "source": [
        "def aget_frames(df, frame_s, hop_s):\n",
        "  N = 3\n",
        "  frames=[]\n",
        "  labels=[]\n",
        "  for i in range(0, len(df) - frame_s, hop_s):\n",
        "    x = df['acc_x'].values[i: i+frame_s]\n",
        "    y = df['acc_y'].values[i: i+frame_s]\n",
        "    z = df['acc_z'].values[i: i+frame_s]\n",
        "\n",
        "    label = stats.mode(df['label'][i:i+frame_s])[0][0]\n",
        "    frames.append([x,y,z])\n",
        "    labels.append(label)\n",
        "  \n",
        "  frames = np.asarray(frames).reshape(-1, frame_s, N)\n",
        "  labels = np.asarray(labels)\n",
        "  \n",
        "  return frames, labels"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "B3gNsg_WgG1n",
        "outputId": "b989e8d2-f4c1-421f-c487-35c671a120b6"
      },
      "source": [
        "AX, AY = aget_frames(Ascaled_X, frame_s, hop_s)\n",
        "AX.shape, AY.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "((4547, 40, 3), (4547,))"
            ]
          },
          "metadata": {},
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4EiASlbZgPu1",
        "outputId": "853844ae-4dcd-4fd3-8543-fd27a07dd644"
      },
      "source": [
        "AX_train, AX_test, AY_train, AY_test = train_test_split(AX, AY, test_size=0.2, stratify=AY)\n",
        "AX_train.shape, AX_test.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "((3637, 40, 3), (910, 40, 3))"
            ]
          },
          "metadata": {},
          "execution_count": 9
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fAQBFiGJgZ__"
      },
      "source": [
        "AX_train = AX_train.reshape(len(AX_train), 40, 3, 1)\n",
        "AX_test = AX_test.reshape(len(AX_test), 40, 3, 1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aJVLsRqegfXM"
      },
      "source": [
        "amodel = Sequential()\n",
        "amodel.add(Conv2D(16, (2,2), activation='relu', input_shape = AX_train[0].shape))\n",
        "amodel.add(Dropout(0.1))\n",
        "\n",
        "amodel.add(Conv2D(32, (2,2), activation='relu'))\n",
        "amodel.add(Dropout(0.2))\n",
        "\n",
        "amodel.add(Flatten())\n",
        "\n",
        "amodel.add(Dense(64, activation='relu'))\n",
        "amodel.add(Dropout(0.5))\n",
        "amodel.add(Dense(9, activation='softmax'))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qZMeAfwDgig3"
      },
      "source": [
        "amodel.compile(optimizer=Adam(learning_rate = 0.001), loss='sparse_categorical_crossentropy', metrics=['accuracy'])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "IbXakrgmgmtP",
        "outputId": "723bd55e-d40b-447a-9c41-938e74eb88de"
      },
      "source": [
        "ahistory = amodel.fit(AX_train, AY_train, epochs=20, validation_data=(AX_test, AY_test), verbose=1)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/20\n",
            "114/114 [==============================] - 1s 8ms/step - loss: 0.5817 - accuracy: 0.8086 - val_loss: 0.1576 - val_accuracy: 0.9505\n",
            "Epoch 2/20\n",
            "114/114 [==============================] - 1s 7ms/step - loss: 0.2109 - accuracy: 0.9315 - val_loss: 0.1145 - val_accuracy: 0.9516\n",
            "Epoch 3/20\n",
            "114/114 [==============================] - 1s 7ms/step - loss: 0.1718 - accuracy: 0.9406 - val_loss: 0.0950 - val_accuracy: 0.9670\n",
            "Epoch 4/20\n",
            "114/114 [==============================] - 1s 7ms/step - loss: 0.1432 - accuracy: 0.9483 - val_loss: 0.0897 - val_accuracy: 0.9593\n",
            "Epoch 5/20\n",
            "114/114 [==============================] - 1s 7ms/step - loss: 0.1362 - accuracy: 0.9533 - val_loss: 0.0960 - val_accuracy: 0.9516\n",
            "Epoch 6/20\n",
            "114/114 [==============================] - 1s 7ms/step - loss: 0.1243 - accuracy: 0.9585 - val_loss: 0.0743 - val_accuracy: 0.9615\n",
            "Epoch 7/20\n",
            "114/114 [==============================] - 1s 7ms/step - loss: 0.1138 - accuracy: 0.9563 - val_loss: 0.0787 - val_accuracy: 0.9747\n",
            "Epoch 8/20\n",
            "114/114 [==============================] - 1s 7ms/step - loss: 0.1075 - accuracy: 0.9621 - val_loss: 0.0891 - val_accuracy: 0.9714\n",
            "Epoch 9/20\n",
            "114/114 [==============================] - 1s 7ms/step - loss: 0.1085 - accuracy: 0.9637 - val_loss: 0.0897 - val_accuracy: 0.9659\n",
            "Epoch 10/20\n",
            "114/114 [==============================] - 1s 7ms/step - loss: 0.0998 - accuracy: 0.9629 - val_loss: 0.0629 - val_accuracy: 0.9780\n",
            "Epoch 11/20\n",
            "114/114 [==============================] - 1s 7ms/step - loss: 0.0934 - accuracy: 0.9665 - val_loss: 0.0611 - val_accuracy: 0.9758\n",
            "Epoch 12/20\n",
            "114/114 [==============================] - 1s 8ms/step - loss: 0.0894 - accuracy: 0.9695 - val_loss: 0.0602 - val_accuracy: 0.9758\n",
            "Epoch 13/20\n",
            "114/114 [==============================] - 1s 7ms/step - loss: 0.0864 - accuracy: 0.9670 - val_loss: 0.0643 - val_accuracy: 0.9791\n",
            "Epoch 14/20\n",
            "114/114 [==============================] - 1s 7ms/step - loss: 0.0893 - accuracy: 0.9692 - val_loss: 0.0598 - val_accuracy: 0.9802\n",
            "Epoch 15/20\n",
            "114/114 [==============================] - 1s 7ms/step - loss: 0.0889 - accuracy: 0.9698 - val_loss: 0.0552 - val_accuracy: 0.9813\n",
            "Epoch 16/20\n",
            "114/114 [==============================] - 1s 7ms/step - loss: 0.0779 - accuracy: 0.9725 - val_loss: 0.0541 - val_accuracy: 0.9802\n",
            "Epoch 17/20\n",
            "114/114 [==============================] - 1s 7ms/step - loss: 0.0798 - accuracy: 0.9714 - val_loss: 0.0527 - val_accuracy: 0.9813\n",
            "Epoch 18/20\n",
            "114/114 [==============================] - 1s 7ms/step - loss: 0.0820 - accuracy: 0.9689 - val_loss: 0.0580 - val_accuracy: 0.9868\n",
            "Epoch 19/20\n",
            "114/114 [==============================] - 1s 8ms/step - loss: 0.0798 - accuracy: 0.9725 - val_loss: 0.0528 - val_accuracy: 0.9824\n",
            "Epoch 20/20\n",
            "114/114 [==============================] - 1s 7ms/step - loss: 0.0795 - accuracy: 0.9687 - val_loss: 0.0515 - val_accuracy: 0.9824\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dM7g-oDpi8cz"
      },
      "source": [
        "def plot_accuracycurve(a, history, epochs):\n",
        "  ep = range(1, epochs+1)\n",
        "  plt.title(a+' Model Accuracy')\n",
        "  plt.plot(ep, history.history['accuracy'])\n",
        "  plt.plot(ep, history.history['val_accuracy'])\n",
        "  plt.ylabel('Accuracy')\n",
        "  plt.ylabel('Epochs')\n",
        "  plt.legend(['Train','Val'])\n",
        "  plt.show()\n",
        "\n",
        "def plot_losscurve(a, history, epochs):\n",
        "  ep = range(1, epochs+1)\n",
        "  plt.title(a+' Model Loss')\n",
        "  plt.plot(ep, history.history['loss'])\n",
        "  plt.plot(ep, history.history['val_loss'])\n",
        "  plt.ylabel('Loss')\n",
        "  plt.ylabel('Epochs')\n",
        "  plt.legend(['Train','Val'])\n",
        "  plt.show()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 545
        },
        "id": "CH4BcH1Ki_5A",
        "outputId": "cfc99060-6733-42ac-f43b-0d0a055b56a3"
      },
      "source": [
        "plot_accuracycurve('Accelerometer',ahistory, 20)\n",
        "plot_losscurve('Accelerometer', ahistory, 20)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qZSzgS6CjC3i"
      },
      "source": [
        "from mlxtend.plotting import plot_confusion_matrix\n",
        "from sklearn.metrics import confusion_matrix"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JHoCK1DgjHkL"
      },
      "source": [
        "AY_P = amodel.predict(AX_test)\n",
        "AY_predict=np.argmax(AY_P,axis=1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 614
        },
        "id": "xtIh8jH-jHqi",
        "outputId": "811b98cd-d4c1-4dd0-bee4-a8ccdb64d764"
      },
      "source": [
        "amat = confusion_matrix(AY_test, AY_predict)\n",
        "plot_confusion_matrix(conf_mat=amat, show_normed=True, colorbar=True, figsize=(10,10), cmap='RdPu')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(<Figure size 720x720 with 2 Axes>,\n",
              " <matplotlib.axes._subplots.AxesSubplot at 0x7f13ab7e2f90>)"
            ]
          },
          "metadata": {},
          "execution_count": 18
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x720 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QSQKEV_yjjrY",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "90c38cc1-c375-45ea-bbd8-084cd8e00d85"
      },
      "source": [
        "print(alabel.classes_)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "['CHAIR' 'LYING_SIDE' 'LYING_UP' 'RUNNING']\n"
          ]
        }
      ]
    }
  ]
}