src/deep_learning.ipynb

{
  "cells": [
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "8UPgncdyTWTU"
      },
      "outputs": [],
      "source": [
        "import tensorflow as tf\n",
        "from tensorflow.keras import datasets, layers, models\n",
        "\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import pandas as pd\n",
        "\n",
        "from sklearn.metrics import classification_report\n",
        "from sklearn.metrics import confusion_matrix, ConfusionMatrixDisplay\n",
        "\n",
        "from classification import Pipeline"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "ZToTFH1cAcIb"
      },
      "outputs": [],
      "source": [
        "pipeline = Pipeline(similarity=True)\n",
        "X_train, y_train, X_valid, y_valid, X_test, y_test = pipeline.load_dataset_from_file().train_valid_test\n",
        "input_shape = X_train.shape[1]\n",
        "\n",
        "print(input_shape)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "LRbMJCvXTajR"
      },
      "outputs": [],
      "source": [
        "model = tf.keras.models.Sequential([\n",
        "    # DNN layers\n",
        "    tf.keras.layers.Dense(input_shape, activation='relu'),\n",
        "    tf.keras.layers.Dense(32, activation='relu'),\n",
        "    tf.keras.layers.Dense(16, activation='relu'),\n",
        "    tf.keras.layers.Dense(8, activation='relu'),\n",
        "    tf.keras.layers.Dense(1, activation='sigmoid'),\n",
        "    tf.keras.layers.Dropout(0.2)\n",
        "])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "IXGC6AnNTaPf"
      },
      "outputs": [],
      "source": [
        "from tensorflow.python.training.saver import metrics\n",
        "from tensorflow.keras.optimizers import Adam\n",
        "\n",
        "optimizer = Adam(learning_rate=0.0001)\n",
        "model.compile(optimizer=optimizer,\n",
        "              loss=tf.keras.losses.BinaryCrossentropy(),\n",
        "              metrics=[\"accuracy\"])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "paYciVVZTl8J",
        "outputId": "800724cc-8bd3-4557-8a0e-9654ab9fe014"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1/150\n",
            "4/4 [==============================] - 2s 80ms/step - loss: 5.5729 - accuracy: 0.5263 - val_loss: 0.5794 - val_accuracy: 0.6000\n",
            "Epoch 2/150\n",
            "4/4 [==============================] - 0s 20ms/step - loss: 2.5783 - accuracy: 0.6316 - val_loss: 0.7184 - val_accuracy: 0.6000\n",
            "Epoch 3/150\n",
            "4/4 [==============================] - 0s 19ms/step - loss: 1.6672 - accuracy: 0.7544 - val_loss: 0.5271 - val_accuracy: 0.7500\n",
            "Epoch 4/150\n",
            "4/4 [==============================] - 0s 26ms/step - loss: 2.4072 - accuracy: 0.7544 - val_loss: 0.5473 - val_accuracy: 0.7500\n",
            "Epoch 5/150\n",
            "4/4 [==============================] - 0s 30ms/step - loss: 3.7578 - accuracy: 0.6491 - val_loss: 0.5913 - val_accuracy: 0.7500\n",
            "Epoch 6/150\n",
            "4/4 [==============================] - 0s 19ms/step - loss: 2.0714 - accuracy: 0.8070 - val_loss: 0.6488 - val_accuracy: 0.7000\n",
            "Epoch 7/150\n",
            "4/4 [==============================] - 0s 25ms/step - loss: 1.9272 - accuracy: 0.7895 - val_loss: 0.8266 - val_accuracy: 0.7000\n",
            "Epoch 8/150\n",
            "4/4 [==============================] - 0s 30ms/step - loss: 4.3304 - accuracy: 0.6491 - val_loss: 1.0329 - val_accuracy: 0.6000\n",
            "Epoch 9/150\n",
            "4/4 [==============================] - 0s 32ms/step - loss: 2.1201 - accuracy: 0.6842 - val_loss: 0.8511 - val_accuracy: 0.6500\n",
            "Epoch 10/150\n",
            "4/4 [==============================] - 0s 23ms/step - loss: 2.7406 - accuracy: 0.7018 - val_loss: 0.5738 - val_accuracy: 0.7000\n",
            "Epoch 11/150\n",
            "4/4 [==============================] - 0s 24ms/step - loss: 1.5914 - accuracy: 0.8421 - val_loss: 0.5423 - val_accuracy: 0.7000\n",
            "Epoch 12/150\n",
            "4/4 [==============================] - 0s 25ms/step - loss: 1.0625 - accuracy: 0.8246 - val_loss: 0.6708 - val_accuracy: 0.6500\n",
            "Epoch 13/150\n",
            "4/4 [==============================] - 0s 28ms/step - loss: 1.6103 - accuracy: 0.7895 - val_loss: 0.7241 - val_accuracy: 0.6500\n",
            "Epoch 14/150\n",
            "4/4 [==============================] - 0s 27ms/step - loss: 3.9936 - accuracy: 0.6842 - val_loss: 0.7313 - val_accuracy: 0.6000\n",
            "Epoch 15/150\n",
            "4/4 [==============================] - 0s 27ms/step - loss: 1.8917 - accuracy: 0.7719 - val_loss: 0.6843 - val_accuracy: 0.6500\n",
            "Epoch 16/150\n",
            "4/4 [==============================] - 0s 25ms/step - loss: 1.3719 - accuracy: 0.7895 - val_loss: 0.5821 - val_accuracy: 0.6500\n",
            "Epoch 17/150\n",
            "4/4 [==============================] - 0s 23ms/step - loss: 2.3512 - accuracy: 0.7895 - val_loss: 0.5090 - val_accuracy: 0.7500\n",
            "Epoch 18/150\n",
            "4/4 [==============================] - 0s 24ms/step - loss: 1.0340 - accuracy: 0.8246 - val_loss: 0.4916 - val_accuracy: 0.8000\n",
            "Epoch 19/150\n",
            "4/4 [==============================] - 0s 25ms/step - loss: 1.8348 - accuracy: 0.8070 - val_loss: 0.4946 - val_accuracy: 0.8000\n",
            "Epoch 20/150\n",
            "4/4 [==============================] - 0s 29ms/step - loss: 2.9124 - accuracy: 0.7544 - val_loss: 0.5201 - val_accuracy: 0.7000\n",
            "Epoch 21/150\n",
            "4/4 [==============================] - 0s 24ms/step - loss: 2.6602 - accuracy: 0.7368 - val_loss: 0.5347 - val_accuracy: 0.7500\n",
            "Epoch 22/150\n",
            "4/4 [==============================] - 0s 24ms/step - loss: 2.0941 - accuracy: 0.7895 - val_loss: 0.5111 - val_accuracy: 0.7500\n",
            "Epoch 23/150\n",
            "4/4 [==============================] - 0s 23ms/step - loss: 1.5753 - accuracy: 0.8070 - val_loss: 0.4902 - val_accuracy: 0.7000\n",
            "Epoch 24/150\n",
            "4/4 [==============================] - 0s 29ms/step - loss: 1.7898 - accuracy: 0.8246 - val_loss: 0.4696 - val_accuracy: 0.7500\n",
            "Epoch 25/150\n",
            "4/4 [==============================] - 0s 34ms/step - loss: 1.8209 - accuracy: 0.7895 - val_loss: 0.4688 - val_accuracy: 0.8000\n",
            "Epoch 26/150\n",
            "4/4 [==============================] - 0s 26ms/step - loss: 2.6261 - accuracy: 0.7368 - val_loss: 0.4713 - val_accuracy: 0.8000\n",
            "Epoch 27/150\n",
            "4/4 [==============================] - 0s 41ms/step - loss: 1.0076 - accuracy: 0.8421 - val_loss: 0.4790 - val_accuracy: 0.8000\n",
            "Epoch 28/150\n",
            "4/4 [==============================] - 0s 36ms/step - loss: 1.0161 - accuracy: 0.8772 - val_loss: 0.4927 - val_accuracy: 0.7500\n",
            "Epoch 29/150\n",
            "4/4 [==============================] - 0s 36ms/step - loss: 0.9858 - accuracy: 0.8947 - val_loss: 0.4807 - val_accuracy: 0.8000\n",
            "Epoch 30/150\n",
            "4/4 [==============================] - 0s 39ms/step - loss: 1.5514 - accuracy: 0.8596 - val_loss: 0.4829 - val_accuracy: 0.7500\n",
            "Epoch 31/150\n",
            "4/4 [==============================] - 0s 32ms/step - loss: 1.5139 - accuracy: 0.8421 - val_loss: 0.5014 - val_accuracy: 0.7500\n",
            "Epoch 32/150\n",
            "4/4 [==============================] - 0s 32ms/step - loss: 2.3111 - accuracy: 0.8246 - val_loss: 0.4978 - val_accuracy: 0.7500\n",
            "Epoch 33/150\n",
            "4/4 [==============================] - 0s 38ms/step - loss: 0.9904 - accuracy: 0.8947 - val_loss: 0.4709 - val_accuracy: 0.7500\n",
            "Epoch 34/150\n",
            "4/4 [==============================] - 0s 42ms/step - loss: 0.9776 - accuracy: 0.9123 - val_loss: 0.4590 - val_accuracy: 0.7500\n",
            "Epoch 35/150\n",
            "4/4 [==============================] - 0s 33ms/step - loss: 3.4000 - accuracy: 0.7544 - val_loss: 0.4678 - val_accuracy: 0.7500\n",
            "Epoch 36/150\n",
            "4/4 [==============================] - 0s 38ms/step - loss: 1.2502 - accuracy: 0.8772 - val_loss: 0.4642 - val_accuracy: 0.7500\n",
            "Epoch 37/150\n",
            "4/4 [==============================] - 0s 38ms/step - loss: 3.1076 - accuracy: 0.7895 - val_loss: 0.4532 - val_accuracy: 0.7500\n",
            "Epoch 38/150\n",
            "4/4 [==============================] - 0s 37ms/step - loss: 2.2957 - accuracy: 0.8421 - val_loss: 0.4487 - val_accuracy: 0.7500\n",
            "Epoch 39/150\n",
            "4/4 [==============================] - 0s 33ms/step - loss: 1.2352 - accuracy: 0.8947 - val_loss: 0.4544 - val_accuracy: 0.7500\n",
            "Epoch 40/150\n",
            "4/4 [==============================] - 0s 41ms/step - loss: 1.7722 - accuracy: 0.8596 - val_loss: 0.4547 - val_accuracy: 0.7500\n",
            "Epoch 41/150\n",
            "4/4 [==============================] - 0s 40ms/step - loss: 1.2213 - accuracy: 0.8947 - val_loss: 0.4445 - val_accuracy: 0.8000\n",
            "Epoch 42/150\n",
            "4/4 [==============================] - 0s 37ms/step - loss: 1.2286 - accuracy: 0.8947 - val_loss: 0.4312 - val_accuracy: 0.8000\n",
            "Epoch 43/150\n",
            "4/4 [==============================] - 0s 34ms/step - loss: 2.5460 - accuracy: 0.8070 - val_loss: 0.4314 - val_accuracy: 0.8000\n",
            "Epoch 44/150\n",
            "4/4 [==============================] - 0s 38ms/step - loss: 3.1128 - accuracy: 0.7719 - val_loss: 0.4558 - val_accuracy: 0.7500\n",
            "Epoch 45/150\n",
            "4/4 [==============================] - 0s 37ms/step - loss: 2.0111 - accuracy: 0.8596 - val_loss: 0.4846 - val_accuracy: 0.7500\n",
            "Epoch 46/150\n",
            "4/4 [==============================] - 0s 40ms/step - loss: 1.7544 - accuracy: 0.8596 - val_loss: 0.4870 - val_accuracy: 0.7500\n",
            "Epoch 47/150\n",
            "4/4 [==============================] - 0s 31ms/step - loss: 2.5277 - accuracy: 0.8246 - val_loss: 0.4952 - val_accuracy: 0.7500\n",
            "Epoch 48/150\n",
            "4/4 [==============================] - 0s 28ms/step - loss: 2.2935 - accuracy: 0.8246 - val_loss: 0.4827 - val_accuracy: 0.7500\n",
            "Epoch 49/150\n",
            "4/4 [==============================] - 0s 23ms/step - loss: 1.9867 - accuracy: 0.8596 - val_loss: 0.4549 - val_accuracy: 0.8000\n",
            "Epoch 50/150\n",
            "4/4 [==============================] - 0s 30ms/step - loss: 1.4859 - accuracy: 0.8772 - val_loss: 0.4730 - val_accuracy: 0.7500\n",
            "Epoch 51/150\n",
            "4/4 [==============================] - 0s 25ms/step - loss: 0.9144 - accuracy: 0.9298 - val_loss: 0.4861 - val_accuracy: 0.7500\n",
            "Epoch 52/150\n",
            "4/4 [==============================] - 0s 24ms/step - loss: 1.1852 - accuracy: 0.9123 - val_loss: 0.4767 - val_accuracy: 0.7500\n",
            "Epoch 53/150\n",
            "4/4 [==============================] - 0s 23ms/step - loss: 2.2742 - accuracy: 0.8246 - val_loss: 0.4643 - val_accuracy: 0.8000\n",
            "Epoch 54/150\n",
            "4/4 [==============================] - 0s 27ms/step - loss: 2.7740 - accuracy: 0.8246 - val_loss: 0.4955 - val_accuracy: 0.7500\n",
            "Epoch 55/150\n",
            "4/4 [==============================] - 0s 28ms/step - loss: 1.7518 - accuracy: 0.8421 - val_loss: 0.5295 - val_accuracy: 0.7500\n",
            "Epoch 56/150\n",
            "4/4 [==============================] - 0s 23ms/step - loss: 1.4312 - accuracy: 0.8947 - val_loss: 0.4541 - val_accuracy: 0.8500\n",
            "Epoch 57/150\n",
            "4/4 [==============================] - 0s 29ms/step - loss: 1.4548 - accuracy: 0.8772 - val_loss: 0.4411 - val_accuracy: 0.8000\n",
            "Epoch 58/150\n",
            "4/4 [==============================] - 0s 23ms/step - loss: 2.5371 - accuracy: 0.8070 - val_loss: 0.4501 - val_accuracy: 0.8500\n",
            "Epoch 59/150\n",
            "4/4 [==============================] - 0s 22ms/step - loss: 1.9815 - accuracy: 0.8596 - val_loss: 0.5392 - val_accuracy: 0.7500\n",
            "Epoch 60/150\n",
            "4/4 [==============================] - 0s 26ms/step - loss: 1.9853 - accuracy: 0.8421 - val_loss: 0.5418 - val_accuracy: 0.7500\n",
            "Epoch 61/150\n",
            "4/4 [==============================] - 0s 22ms/step - loss: 1.9932 - accuracy: 0.8421 - val_loss: 0.4961 - val_accuracy: 0.8000\n",
            "Epoch 62/150\n",
            "4/4 [==============================] - 0s 20ms/step - loss: 1.1746 - accuracy: 0.8947 - val_loss: 0.4349 - val_accuracy: 0.8000\n",
            "Epoch 63/150\n",
            "4/4 [==============================] - 0s 19ms/step - loss: 1.7353 - accuracy: 0.8772 - val_loss: 0.4378 - val_accuracy: 0.8500\n",
            "Epoch 64/150\n",
            "4/4 [==============================] - 0s 25ms/step - loss: 2.2534 - accuracy: 0.8421 - val_loss: 0.4869 - val_accuracy: 0.7500\n",
            "Epoch 65/150\n",
            "4/4 [==============================] - 0s 20ms/step - loss: 2.7840 - accuracy: 0.8070 - val_loss: 0.5277 - val_accuracy: 0.7500\n",
            "Epoch 66/150\n",
            "4/4 [==============================] - 0s 20ms/step - loss: 3.0496 - accuracy: 0.7895 - val_loss: 0.5247 - val_accuracy: 0.7500\n",
            "Epoch 67/150\n",
            "4/4 [==============================] - 0s 19ms/step - loss: 1.9838 - accuracy: 0.8596 - val_loss: 0.5177 - val_accuracy: 0.7500\n",
            "Epoch 68/150\n",
            "4/4 [==============================] - 0s 26ms/step - loss: 1.4447 - accuracy: 0.8772 - val_loss: 0.4788 - val_accuracy: 0.8000\n",
            "Epoch 69/150\n",
            "4/4 [==============================] - 0s 20ms/step - loss: 1.4323 - accuracy: 0.8772 - val_loss: 0.4462 - val_accuracy: 0.8500\n",
            "Epoch 70/150\n",
            "4/4 [==============================] - 0s 19ms/step - loss: 1.1684 - accuracy: 0.8947 - val_loss: 0.4784 - val_accuracy: 0.8000\n",
            "Epoch 71/150\n",
            "4/4 [==============================] - 0s 20ms/step - loss: 1.9772 - accuracy: 0.8596 - val_loss: 0.4916 - val_accuracy: 0.8000\n",
            "Epoch 72/150\n",
            "4/4 [==============================] - 0s 22ms/step - loss: 1.1678 - accuracy: 0.9123 - val_loss: 0.5043 - val_accuracy: 0.8000\n",
            "Epoch 73/150\n",
            "4/4 [==============================] - 0s 26ms/step - loss: 1.9599 - accuracy: 0.8596 - val_loss: 0.5227 - val_accuracy: 0.7500\n",
            "Epoch 74/150\n",
            "4/4 [==============================] - 0s 31ms/step - loss: 2.2381 - accuracy: 0.8246 - val_loss: 0.5079 - val_accuracy: 0.8000\n",
            "Epoch 75/150\n",
            "4/4 [==============================] - 0s 29ms/step - loss: 1.4276 - accuracy: 0.8772 - val_loss: 0.4702 - val_accuracy: 0.8000\n",
            "Epoch 76/150\n",
            "4/4 [==============================] - 0s 30ms/step - loss: 1.4233 - accuracy: 0.8947 - val_loss: 0.4821 - val_accuracy: 0.8000\n",
            "Epoch 77/150\n",
            "4/4 [==============================] - 0s 23ms/step - loss: 2.2066 - accuracy: 0.8596 - val_loss: 0.5152 - val_accuracy: 0.8000\n",
            "Epoch 78/150\n",
            "4/4 [==============================] - 0s 27ms/step - loss: 1.6868 - accuracy: 0.8772 - val_loss: 0.5329 - val_accuracy: 0.7500\n",
            "Epoch 79/150\n",
            "4/4 [==============================] - 0s 32ms/step - loss: 2.2290 - accuracy: 0.8421 - val_loss: 0.5109 - val_accuracy: 0.8000\n",
            "Epoch 80/150\n",
            "4/4 [==============================] - 0s 26ms/step - loss: 1.1437 - accuracy: 0.9123 - val_loss: 0.4965 - val_accuracy: 0.8000\n",
            "Epoch 81/150\n",
            "4/4 [==============================] - 0s 24ms/step - loss: 1.9559 - accuracy: 0.8596 - val_loss: 0.4736 - val_accuracy: 0.8000\n",
            "Epoch 82/150\n",
            "4/4 [==============================] - 0s 29ms/step - loss: 1.6903 - accuracy: 0.8772 - val_loss: 0.5135 - val_accuracy: 0.8000\n",
            "Epoch 83/150\n",
            "4/4 [==============================] - 0s 25ms/step - loss: 1.9881 - accuracy: 0.8421 - val_loss: 0.5372 - val_accuracy: 0.7500\n",
            "Epoch 84/150\n",
            "4/4 [==============================] - 0s 26ms/step - loss: 2.2411 - accuracy: 0.8246 - val_loss: 0.4802 - val_accuracy: 0.8500\n",
            "Epoch 85/150\n",
            "4/4 [==============================] - 0s 24ms/step - loss: 2.2209 - accuracy: 0.8596 - val_loss: 0.4403 - val_accuracy: 0.8500\n",
            "Epoch 86/150\n",
            "4/4 [==============================] - 0s 24ms/step - loss: 2.2432 - accuracy: 0.8596 - val_loss: 0.4772 - val_accuracy: 0.8000\n",
            "Epoch 87/150\n",
            "4/4 [==============================] - 0s 25ms/step - loss: 2.7612 - accuracy: 0.8246 - val_loss: 0.5335 - val_accuracy: 0.8000\n",
            "Epoch 88/150\n",
            "4/4 [==============================] - 0s 28ms/step - loss: 1.1506 - accuracy: 0.8947 - val_loss: 0.5612 - val_accuracy: 0.7500\n",
            "Epoch 89/150\n",
            "4/4 [==============================] - 0s 27ms/step - loss: 2.5069 - accuracy: 0.8070 - val_loss: 0.4955 - val_accuracy: 0.8000\n",
            "Epoch 90/150\n",
            "4/4 [==============================] - 0s 24ms/step - loss: 1.9586 - accuracy: 0.8596 - val_loss: 0.4843 - val_accuracy: 0.8000\n",
            "Epoch 91/150\n",
            "4/4 [==============================] - 0s 27ms/step - loss: 4.3836 - accuracy: 0.7018 - val_loss: 0.5168 - val_accuracy: 0.8000\n",
            "Epoch 92/150\n",
            "4/4 [==============================] - 0s 25ms/step - loss: 1.9548 - accuracy: 0.8596 - val_loss: 0.5732 - val_accuracy: 0.7500\n",
            "Epoch 93/150\n",
            "4/4 [==============================] - 0s 24ms/step - loss: 1.9430 - accuracy: 0.8596 - val_loss: 0.5217 - val_accuracy: 0.8000\n",
            "Epoch 94/150\n",
            "4/4 [==============================] - 0s 29ms/step - loss: 3.0230 - accuracy: 0.8070 - val_loss: 0.5037 - val_accuracy: 0.8000\n",
            "Epoch 95/150\n",
            "4/4 [==============================] - 0s 23ms/step - loss: 1.9528 - accuracy: 0.8421 - val_loss: 0.5161 - val_accuracy: 0.8000\n",
            "Epoch 96/150\n",
            "4/4 [==============================] - 0s 27ms/step - loss: 2.7472 - accuracy: 0.8246 - val_loss: 0.5255 - val_accuracy: 0.8000\n",
            "Epoch 97/150\n",
            "4/4 [==============================] - 0s 23ms/step - loss: 2.2172 - accuracy: 0.8421 - val_loss: 0.5092 - val_accuracy: 0.8000\n",
            "Epoch 98/150\n",
            "4/4 [==============================] - 0s 27ms/step - loss: 0.8639 - accuracy: 0.9474 - val_loss: 0.4938 - val_accuracy: 0.8000\n",
            "Epoch 99/150\n",
            "4/4 [==============================] - 0s 24ms/step - loss: 1.4066 - accuracy: 0.9123 - val_loss: 0.5116 - val_accuracy: 0.8000\n",
            "Epoch 100/150\n",
            "4/4 [==============================] - 0s 30ms/step - loss: 1.1172 - accuracy: 0.9298 - val_loss: 0.5390 - val_accuracy: 0.8000\n",
            "Epoch 101/150\n",
            "4/4 [==============================] - 0s 23ms/step - loss: 2.4956 - accuracy: 0.8246 - val_loss: 0.5623 - val_accuracy: 0.8000\n",
            "Epoch 102/150\n",
            "4/4 [==============================] - 0s 24ms/step - loss: 1.1476 - accuracy: 0.9123 - val_loss: 0.4884 - val_accuracy: 0.8500\n",
            "Epoch 103/150\n",
            "4/4 [==============================] - 0s 32ms/step - loss: 2.7465 - accuracy: 0.8246 - val_loss: 0.5031 - val_accuracy: 0.8500\n",
            "Epoch 104/150\n",
            "4/4 [==============================] - 0s 23ms/step - loss: 2.4767 - accuracy: 0.8421 - val_loss: 0.5463 - val_accuracy: 0.8000\n",
            "Epoch 105/150\n",
            "4/4 [==============================] - 0s 24ms/step - loss: 1.6716 - accuracy: 0.8772 - val_loss: 0.5323 - val_accuracy: 0.8000\n",
            "Epoch 106/150\n",
            "4/4 [==============================] - 0s 26ms/step - loss: 2.4691 - accuracy: 0.8246 - val_loss: 0.5221 - val_accuracy: 0.8000\n",
            "Epoch 107/150\n",
            "4/4 [==============================] - 0s 27ms/step - loss: 0.5785 - accuracy: 0.9649 - val_loss: 0.5062 - val_accuracy: 0.8000\n",
            "Epoch 108/150\n",
            "4/4 [==============================] - 0s 27ms/step - loss: 1.1311 - accuracy: 0.9298 - val_loss: 0.5207 - val_accuracy: 0.8000\n",
            "Epoch 109/150\n",
            "4/4 [==============================] - 0s 28ms/step - loss: 1.9322 - accuracy: 0.8772 - val_loss: 0.5362 - val_accuracy: 0.8000\n",
            "Epoch 110/150\n",
            "4/4 [==============================] - 0s 23ms/step - loss: 1.3923 - accuracy: 0.9123 - val_loss: 0.5511 - val_accuracy: 0.8000\n",
            "Epoch 111/150\n",
            "4/4 [==============================] - 0s 24ms/step - loss: 3.2818 - accuracy: 0.7895 - val_loss: 0.5845 - val_accuracy: 0.8000\n",
            "Epoch 112/150\n",
            "4/4 [==============================] - 0s 24ms/step - loss: 2.7471 - accuracy: 0.8070 - val_loss: 0.5804 - val_accuracy: 0.8000\n",
            "Epoch 113/150\n",
            "4/4 [==============================] - 0s 29ms/step - loss: 2.2049 - accuracy: 0.8596 - val_loss: 0.5360 - val_accuracy: 0.8000\n",
            "Epoch 114/150\n",
            "4/4 [==============================] - 0s 29ms/step - loss: 1.1286 - accuracy: 0.9298 - val_loss: 0.5364 - val_accuracy: 0.8000\n",
            "Epoch 115/150\n",
            "4/4 [==============================] - 0s 23ms/step - loss: 1.6466 - accuracy: 0.8947 - val_loss: 0.6270 - val_accuracy: 0.7500\n",
            "Epoch 116/150\n",
            "4/4 [==============================] - 0s 25ms/step - loss: 1.6592 - accuracy: 0.8772 - val_loss: 0.6190 - val_accuracy: 0.7500\n",
            "Epoch 117/150\n",
            "4/4 [==============================] - 0s 23ms/step - loss: 2.2048 - accuracy: 0.8596 - val_loss: 0.5393 - val_accuracy: 0.8000\n",
            "Epoch 118/150\n",
            "4/4 [==============================] - 0s 23ms/step - loss: 2.4757 - accuracy: 0.8421 - val_loss: 0.4878 - val_accuracy: 0.8500\n",
            "Epoch 119/150\n",
            "4/4 [==============================] - 0s 25ms/step - loss: 2.2061 - accuracy: 0.8596 - val_loss: 0.5608 - val_accuracy: 0.8000\n",
            "Epoch 120/150\n",
            "4/4 [==============================] - 0s 30ms/step - loss: 1.6485 - accuracy: 0.8947 - val_loss: 0.6452 - val_accuracy: 0.7500\n",
            "Epoch 121/150\n",
            "4/4 [==============================] - 0s 24ms/step - loss: 1.6585 - accuracy: 0.8772 - val_loss: 0.5964 - val_accuracy: 0.8000\n",
            "Epoch 122/150\n",
            "4/4 [==============================] - 0s 28ms/step - loss: 1.9294 - accuracy: 0.8772 - val_loss: 0.5434 - val_accuracy: 0.8000\n",
            "Epoch 123/150\n",
            "4/4 [==============================] - 0s 27ms/step - loss: 1.9211 - accuracy: 0.8772 - val_loss: 0.5439 - val_accuracy: 0.8000\n",
            "Epoch 124/150\n",
            "4/4 [==============================] - 0s 23ms/step - loss: 0.8459 - accuracy: 0.9474 - val_loss: 0.5719 - val_accuracy: 0.8000\n",
            "Epoch 125/150\n",
            "4/4 [==============================] - 0s 23ms/step - loss: 0.8520 - accuracy: 0.9474 - val_loss: 0.6301 - val_accuracy: 0.8000\n",
            "Epoch 126/150\n",
            "4/4 [==============================] - 0s 30ms/step - loss: 0.3048 - accuracy: 0.9825 - val_loss: 0.5434 - val_accuracy: 0.8000\n",
            "Epoch 127/150\n",
            "4/4 [==============================] - 0s 22ms/step - loss: 2.2008 - accuracy: 0.8596 - val_loss: 0.5698 - val_accuracy: 0.8000\n",
            "Epoch 128/150\n",
            "4/4 [==============================] - 0s 23ms/step - loss: 0.5620 - accuracy: 0.9649 - val_loss: 0.5969 - val_accuracy: 0.8000\n",
            "Epoch 129/150\n",
            "4/4 [==============================] - 0s 27ms/step - loss: 1.3757 - accuracy: 0.9123 - val_loss: 0.6140 - val_accuracy: 0.8000\n",
            "Epoch 130/150\n",
            "4/4 [==============================] - 0s 24ms/step - loss: 1.3794 - accuracy: 0.9123 - val_loss: 0.5785 - val_accuracy: 0.8000\n",
            "Epoch 131/150\n",
            "4/4 [==============================] - 0s 25ms/step - loss: 2.4601 - accuracy: 0.8421 - val_loss: 0.5574 - val_accuracy: 0.8000\n",
            "Epoch 132/150\n",
            "4/4 [==============================] - 0s 28ms/step - loss: 1.0988 - accuracy: 0.9298 - val_loss: 0.5901 - val_accuracy: 0.8000\n",
            "Epoch 133/150\n",
            "4/4 [==============================] - 0s 27ms/step - loss: 1.9196 - accuracy: 0.8772 - val_loss: 0.5934 - val_accuracy: 0.8000\n",
            "Epoch 134/150\n",
            "4/4 [==============================] - 0s 20ms/step - loss: 2.4583 - accuracy: 0.8421 - val_loss: 0.6203 - val_accuracy: 0.8000\n",
            "Epoch 135/150\n",
            "4/4 [==============================] - 0s 21ms/step - loss: 2.4497 - accuracy: 0.8421 - val_loss: 0.6215 - val_accuracy: 0.8000\n",
            "Epoch 136/150\n",
            "4/4 [==============================] - 0s 22ms/step - loss: 0.8274 - accuracy: 0.9474 - val_loss: 0.5739 - val_accuracy: 0.8500\n",
            "Epoch 137/150\n",
            "4/4 [==============================] - 0s 21ms/step - loss: 1.3701 - accuracy: 0.9123 - val_loss: 0.5726 - val_accuracy: 0.8500\n",
            "Epoch 138/150\n",
            "4/4 [==============================] - 0s 20ms/step - loss: 1.6433 - accuracy: 0.8947 - val_loss: 0.6131 - val_accuracy: 0.8000\n",
            "Epoch 139/150\n",
            "4/4 [==============================] - 0s 28ms/step - loss: 2.1850 - accuracy: 0.8596 - val_loss: 0.6561 - val_accuracy: 0.8000\n",
            "Epoch 140/150\n",
            "4/4 [==============================] - 0s 22ms/step - loss: 2.1813 - accuracy: 0.8596 - val_loss: 0.6031 - val_accuracy: 0.8000\n",
            "Epoch 141/150\n",
            "4/4 [==============================] - 0s 21ms/step - loss: 1.6447 - accuracy: 0.8947 - val_loss: 0.5843 - val_accuracy: 0.8000\n",
            "Epoch 142/150\n",
            "4/4 [==============================] - 0s 25ms/step - loss: 1.9102 - accuracy: 0.8772 - val_loss: 0.6170 - val_accuracy: 0.8000\n",
            "Epoch 143/150\n",
            "4/4 [==============================] - 0s 21ms/step - loss: 2.7218 - accuracy: 0.8246 - val_loss: 0.6156 - val_accuracy: 0.8000\n",
            "Epoch 144/150\n",
            "4/4 [==============================] - 0s 24ms/step - loss: 2.4489 - accuracy: 0.8421 - val_loss: 0.6135 - val_accuracy: 0.8000\n",
            "Epoch 145/150\n",
            "4/4 [==============================] - 0s 24ms/step - loss: 1.0946 - accuracy: 0.9298 - val_loss: 0.6166 - val_accuracy: 0.8000\n",
            "Epoch 146/150\n",
            "4/4 [==============================] - 0s 26ms/step - loss: 1.3666 - accuracy: 0.9123 - val_loss: 0.6048 - val_accuracy: 0.8000\n",
            "Epoch 147/150\n",
            "4/4 [==============================] - 0s 21ms/step - loss: 1.9024 - accuracy: 0.8772 - val_loss: 0.6411 - val_accuracy: 0.8000\n",
            "Epoch 148/150\n",
            "4/4 [==============================] - 0s 26ms/step - loss: 0.5585 - accuracy: 0.9649 - val_loss: 0.6473 - val_accuracy: 0.8000\n",
            "Epoch 149/150\n",
            "4/4 [==============================] - 0s 25ms/step - loss: 2.1767 - accuracy: 0.8596 - val_loss: 0.6186 - val_accuracy: 0.8000\n",
            "Epoch 150/150\n",
            "4/4 [==============================] - 0s 35ms/step - loss: 2.9869 - accuracy: 0.8070 - val_loss: 0.5855 - val_accuracy: 0.8500\n"
          ]
        }
      ],
      "source": [
        "NUM_EPOCHS = 150\n",
        "\n",
        "history = model.fit(\n",
        "      x = X_train,\n",
        "      y = y_train,\n",
        "      epochs = NUM_EPOCHS,\n",
        "      batch_size=16,\n",
        "      verbose = 1,\n",
        "      validation_data=(X_valid, y_valid))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "R0RDeOQKDFbq",
        "outputId": "345e2fb9-0dc3-460b-eafc-2e65a65b1fe9"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "1/1 [==============================] - 0s 35ms/step - loss: 0.4795 - accuracy: 0.9000\n"
          ]
        },
        {
          "data": {
            "text/plain": [
              "[0.47951340675354004, 0.8999999761581421]"
            ]
          },
          "execution_count": 100,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "model.evaluate(X_test, y_test)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "5-zwuYAgCo8N",
        "outputId": "42374027-7b5a-424a-f1b8-068495bd3779"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "1/1 [==============================] - 0s 83ms/step\n",
            "[1 0 1 1 1 1 1 1 1 1 1 0 0 1 0 0 0 0 0 0]\n",
            "[1 0 1 0 1 1 1 1 1 1 1 0 0 1 0 1 0 0 0 0]\n"
          ]
        }
      ],
      "source": [
        "predictions = []\n",
        "for value in (model.predict(X_test)>=0.5).astype(int):\n",
        "  predictions.append(value[0])\n",
        "predictions = np.array(predictions)\n",
        "actual = y_test\n",
        "\n",
        "print(predictions)\n",
        "print(actual)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 448
        },
        "id": "v-2a7j-pTl5y",
        "outputId": "93eb597b-85c7-4852-b4a3-0c2f97fb5ab1"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "def moving_average(x, w):\n",
        "    return np.convolve(x, np.ones(w), 'valid') / w\n",
        "\n",
        "AVERAGING = 20\n",
        "\n",
        "train_accs = history.history['accuracy']\n",
        "train_graph = moving_average(train_accs, AVERAGING)\n",
        "epoch_range = [x + 1 for x in range(len(train_graph))]\n",
        "\n",
        "validation_accs = history.history['val_accuracy']\n",
        "val_graph = moving_average(validation_accs, AVERAGING)\n",
        "epoch_range = [x + 1 for x in range(len(val_graph))]\n",
        "\n",
        "plt.plot(np.asarray(epoch_range).astype(str), train_graph, label='Training Accuracy')\n",
        "plt.plot(np.asarray(epoch_range).astype(str), val_graph, label='Validation Accuracy')\n",
        "plt.xticks([])\n",
        "plt.title('Fake News Classifier with BERT')\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.legend(loc='lower right')\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_nO9xA3Zpm0k",
        "outputId": "445decfa-0a84-4e7d-fe06-352a5d1a396e"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "1/1 [==============================] - 0s 32ms/step\n",
            "0.9090909090909091\n",
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.89      0.89      0.89         9\n",
            "           1       0.91      0.91      0.91        11\n",
            "\n",
            "    accuracy                           0.90        20\n",
            "   macro avg       0.90      0.90      0.90        20\n",
            "weighted avg       0.90      0.90      0.90        20\n",
            "\n"
          ]
        }
      ],
      "source": [
        "predictions = []\n",
        "for value in (model.predict(X_test)>=0.5).astype(int):\n",
        "  predictions.append(value[0])\n",
        "predictions = np.array(predictions)\n",
        "actual = y_test\n",
        "\n",
        "print(classification_report(actual, predictions))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 455
        },
        "id": "kw0g9lH2qfGP",
        "outputId": "2eb88d32-6eb9-4adc-f630-ca49ba6531ab"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 2 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "# confusion matrix\n",
        "cm = confusion_matrix(actual, predictions, labels=[0,1])\n",
        "disp = ConfusionMatrixDisplay(confusion_matrix=cm, display_labels=[\"Fake\", \"Real\"])\n",
        "disp.plot()\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "MWm0LBLCDq7q"
      },
      "outputs": [],
      "source": [
        "#model.save(\"fake_news_classification_model\")"
      ]
    }
  ],
  "metadata": {
    "colab": {
      "provenance": []
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}