email_classification.ipynb

{
  "cells": [
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "id": "5Dlr7aI4m4xu"
      },
      "outputs": [],
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.preprocessing import MinMaxScaler\n",
        "from sklearn.metrics import classification_report\n",
        "\n",
        "import tensorflow as tf\n",
        "from tensorflow import keras\n",
        "\n",
        "import keras_tuner as kt\n",
        "\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "import io\n",
        "\n",
        "import matplotlib.pyplot as plt"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {
        "id": "0mJ1WQAbpkB3"
      },
      "outputs": [],
      "source": [
        "emails = pd.read_csv(\"./emails.csv\")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9FacRejBIXpv"
      },
      "source": [
        "In this project, we build a \"Spam Classifier\". Our dataset consists of 5172 \n",
        "examples and 3002 features. We have 1500 \"Spam\" emails and 3672 \"Ham\". The feature in the index 0 is a number id for each email (not important). The last feature is the class for each email (1 is Spam, 0 is Ham). The rest of the features are words' counts in each email. "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {
        "id": "tbtT8FTcrc-Q"
      },
      "outputs": [],
      "source": [
        "X = emails.drop(columns=[\"Email No.\", \"Prediction\"])\n",
        "y = emails[[\"Prediction\"]]"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "YJ-lvF6_Lw__"
      },
      "source": [
        "# Create Train, Develop and Test Set\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "aJUzlso7MGUH"
      },
      "source": [
        "We split our dataset into 70% training set, 20% for development, and 10% to test set. The final model is evaluated against the test set."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "id": "gan13nXlxW86"
      },
      "outputs": [],
      "source": [
        "X_train, X_dev, y_train, y_dev = train_test_split(X, y, test_size=0.25, \n",
        "                                                    random_state=42, \n",
        "                                                    shuffle=True)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "metadata": {
        "id": "171FY5SZDY10"
      },
      "outputs": [],
      "source": [
        "X_dev, X_test, y_dev, y_test = train_test_split(X_dev, y_dev, test_size=0.3,\n",
        "                                                random_state=42)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 10,
      "metadata": {
        "id": "l0c3apj5wZvZ"
      },
      "outputs": [],
      "source": [
        "scaler = MinMaxScaler()\n",
        "X_train = scaler.fit_transform(X_train)\n",
        "X_dev = scaler.transform(X_dev)\n",
        "X_test = scaler.transform(X_test)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "metadata": {
        "id": "MQKNt7w82me1"
      },
      "outputs": [],
      "source": [
        "# Helper function to create plots\n",
        "def plot_graphs(history, metric):\n",
        "  plt.plot(history.history[metric])\n",
        "  plt.plot(history.history['val_'+metric], '')\n",
        "  plt.xlabel(\"Epochs\")\n",
        "  plt.ylabel(metric)\n",
        "  plt.legend([metric, 'val_'+metric])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 48,
      "metadata": {
        "id": "qQxtRsvoDvev"
      },
      "outputs": [],
      "source": [
        "# report is a dictionary to keep track of all models\n",
        "# and the corresponded f1score on SPAM class.\n",
        "report = {}"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2z_VvpVZMnna"
      },
      "source": [
        "# Base Network\n",
        "\n",
        "Our base network model has two hidden and an output layer (this architect is totally random).\n",
        "\n",
        "We are using \"MinMax Normalization\" to normalize our inputs. Thus, all inputs are in the range of (0,1). \n",
        "\n",
        "Next, we have two \"Dense\" layers with 128 and 256 neurons respectively. Because we use the \"ReLU\" activation function for our hidden layers we also use the \"He normal\" class to initialize our weights. This way we prevent our network from suffering from the vanishing/exploding gradients.\n",
        "\n",
        "For the output layer, we choose the \"sigmoid\" activation function because our problem is the binary meaning we want to output just two classes (0 or 1)."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 11,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "M81iBccwSRZE",
        "outputId": "e972feb7-fdb9-4734-f9bf-cb43c034ee66"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " dense (Dense)               (None, 128)               384128    \n",
            "                                                                 \n",
            " dense_1 (Dense)             (None, 256)               33024     \n",
            "                                                                 \n",
            " dense_2 (Dense)             (None, 1)                 257       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 417,409\n",
            "Trainable params: 417,409\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ],
      "source": [
        "model = keras.models.Sequential([\n",
        "    keras.layers.InputLayer(input_shape=(3000,)),\n",
        "    keras.layers.Dense(128, activation=\"relu\", kernel_initializer=\"he_normal\"),\n",
        "    keras.layers.Dense(256, activation=\"relu\", kernel_initializer=\"he_normal\"),\n",
        "    keras.layers.Dense(1, activation=\"sigmoid\")\n",
        "])\n",
        "\n",
        "model.summary()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "sGYVD0tt496S"
      },
      "source": [
        "We pick \"Adam\" optimizer with a default configuration to optimize the \"Binary Cross Entropy\" (or Log) loss function. Adam optimizer is a combination of the \"momentum\" and \"RMSprop\" optimization techniques. Thus, it keeps track of an exponentially decaying average of past gradients and an exponentially decaying average of past squared gradients.\n",
        "\n",
        "When fitting the model we keep track of the \"accuracy\", \"recall\" and \"precision\" metrics to evaluate our base network against more complex neural networks."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "RarRBCg0S3pk"
      },
      "outputs": [],
      "source": [
        "optimizer = keras.optimizers.Adam()\n",
        "\n",
        "model.compile(loss=keras.losses.BinaryCrossentropy(),\n",
        "              optimizer=optimizer,\n",
        "              metrics=[\n",
        "                tf.keras.metrics.BinaryAccuracy(name='accuracy'),\n",
        "                tf.keras.metrics.BinaryAccuracy(name='recall'),\n",
        "                tf.keras.metrics.BinaryAccuracy(name='precision')\n",
        "              ])\n",
        "\n",
        "history = model.fit(X_train, np.array(y_train), epochs=10,\n",
        "                    validation_data=(X_dev, np.array(y_dev)))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 13,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 518
        },
        "id": "qOMfA7UQ2ai_",
        "outputId": "e903201a-a181-4504-e1f8-828ba8e29d2a"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "(0.0, 0.3269346948014572)"
            ]
          },
          "execution_count": 13,
          "metadata": {},
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 1152x576 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "plt.figure(figsize=(16, 8))\n",
        "plt.subplot(1, 2, 1)\n",
        "plot_graphs(history, 'accuracy')\n",
        "plt.ylim(None, 1)\n",
        "plt.subplot(1, 2, 2)\n",
        "plot_graphs(history, 'loss')\n",
        "plt.ylim(0, None)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bNO5Pxxj93ft"
      },
      "source": [
        "Based on the above plots our network seems to overfit our dataset. The loss curve on the development dataset starts to increase and on the 6th epoch we have a pick. Then, the network turns back to $0.1$ but still the **validation loss** remains almost 10 times larger than **train loss**."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 44,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Hhf5UHoimhe4",
        "outputId": "64ca882f-0b9e-4b90-9d73-718f1795e72d"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "         ham       0.99      0.96      0.97       270\n",
            "        spam       0.91      0.97      0.94       118\n",
            "\n",
            "    accuracy                           0.96       388\n",
            "   macro avg       0.95      0.97      0.96       388\n",
            "weighted avg       0.96      0.96      0.96       388\n",
            "\n"
          ]
        }
      ],
      "source": [
        "y_pred = model.predict(X_test) > 0.5\n",
        "\n",
        "target_names = [\"ham\", \"spam\"]\n",
        "print(classification_report(np.array(y_test), y_pred, target_names=target_names))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "GdUgPQ8a_bKZ"
      },
      "source": [
        "We focus on the \"Spam\" class. Our network has a very high recall, meaning that identifies correctly 97% of the true \"Spam\" emails. The precision score is above 90% but a bit lower than recall. Precision equals to $.91$ means that our classifier predicts a **ham** email as **spam** almost 1 of 10 times."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 54,
      "metadata": {
        "id": "DSkV58QDD89Q"
      },
      "outputs": [],
      "source": [
        "report[\"base_model\"] = classification_report(np.array(y_test), y_pred, target_names=target_names, output_dict=True)[\"spam\"][\"f1-score\"]"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "SAIYg6qSHBqH"
      },
      "source": [
        "# Model 1 - Fine Tuning\n",
        "\n",
        "The learning rate hyperparameter is one of the most significant we need to fine-tune as it handles the optimization of our network. The learning rate is responsible for how fast the network will converge to a global minimum. A too-large learning rate is possible for the network to not converge whereas too small a learning rate is possible to make the training process slow.\n",
        "\n",
        "In addition, we fine-tune the **# of hidden units** per layer."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 15,
      "metadata": {
        "id": "C08pZGv1yzOj"
      },
      "outputs": [],
      "source": [
        "def build_model(hp):\n",
        "\n",
        "  # Tune the number of units in the first Dense layer\n",
        "  # Choose an optimal value between 32-512\n",
        "  hp_units1 = hp.Int('units_1', min_value=32, max_value=512, step=32)\n",
        "  hp_units2 = hp.Int('units_2', min_value=32, max_value=512, step=32)\n",
        "\n",
        "  model = keras.models.Sequential([\n",
        "      keras.layers.InputLayer(input_shape=(3000,)),\n",
        "      keras.layers.Dense(hp_units1, activation=\"relu\", kernel_initializer=\"he_normal\"),\n",
        "      keras.layers.Dense(hp_units2, activation=\"relu\", kernel_initializer=\"he_normal\"),\n",
        "      keras.layers.Dense(1, activation=\"sigmoid\")\n",
        "  ])\n",
        "\n",
        "  # Tune the learning rate for the optimizer\n",
        "  # Choose an optimal value from 0.01, 0.001, or 0.0001\n",
        "  hp_learning_rate = hp.Choice('learning_rate', values=[1e-2, 1e-3, 1e-4])\n",
        "\n",
        "  optimizer = keras.optimizers.Adam(learning_rate=hp_learning_rate)\n",
        "\n",
        "  model.compile(loss=keras.losses.BinaryCrossentropy(),\n",
        "              optimizer=optimizer,\n",
        "              metrics=[\n",
        "                tf.keras.metrics.BinaryAccuracy(name='accuracy'),\n",
        "                tf.keras.metrics.BinaryAccuracy(name='precision')\n",
        "              ])\n",
        "  \n",
        "  return model"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 16,
      "metadata": {
        "id": "DxPTeMjH0r98"
      },
      "outputs": [],
      "source": [
        "tuner = kt.Hyperband(build_model,\n",
        "                     objective=kt.Objective(\"val_precision\", direction=\"max\"),\n",
        "                     max_epochs=10,\n",
        "                     factor=3,\n",
        "                     directory='my_dir',\n",
        "                     project_name='email_classification')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 17,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "795EuSJVtFLb",
        "outputId": "6edb3930-de87-4bbf-f188-89483249677a"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Trial 30 Complete [00h 00m 07s]\n",
            "val_precision: 0.9878453016281128\n",
            "\n",
            "Best val_precision So Far: 0.9878453016281128\n",
            "Total elapsed time: 00h 03m 31s\n",
            "INFO:tensorflow:Oracle triggered exit\n"
          ]
        }
      ],
      "source": [
        "tuner.search(X_train, np.array(y_train), epochs=10,\n",
        "                    validation_data=(X_dev, np.array(y_dev)))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 18,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "KrHnVZrTb1kV",
        "outputId": "8ce97bb1-ee17-4d34-fe00-908b80f455ac"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\n",
            "The hyperparameter search is complete. The optimal number of units in the first densely-connected\n",
            "layer is 64, the optimal number of units in the second densely-connected\n",
            "layer is 320 and the optimal learning rate for the optimizer\n",
            "is 0.001.\n",
            "\n"
          ]
        }
      ],
      "source": [
        "# Get the optimal hyperparameters\n",
        "best_hps=tuner.get_best_hyperparameters(num_trials=30)[0]\n",
        "\n",
        "print(f\"\"\"\n",
        "The hyperparameter search is complete. The optimal number of units in the first densely-connected\n",
        "layer is {best_hps.get('units_1')}, the optimal number of units in the second densely-connected\n",
        "layer is {best_hps.get('units_2')} and the optimal learning rate for the optimizer\n",
        "is {best_hps.get('learning_rate')}.\n",
        "\"\"\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 19,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "KwStrqjeenr0",
        "outputId": "bc88e0e7-f10e-47bd-fa5b-76a1e4fac04c"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1/10\n",
            "122/122 [==============================] - 2s 8ms/step - loss: 0.2708 - accuracy: 0.8786 - precision: 0.8786 - val_loss: 0.0822 - val_accuracy: 0.9779 - val_precision: 0.9779\n",
            "Epoch 2/10\n",
            "122/122 [==============================] - 1s 5ms/step - loss: 0.0464 - accuracy: 0.9874 - precision: 0.9874 - val_loss: 0.0876 - val_accuracy: 0.9757 - val_precision: 0.9757\n",
            "Epoch 3/10\n",
            "122/122 [==============================] - 1s 6ms/step - loss: 0.0245 - accuracy: 0.9943 - precision: 0.9943 - val_loss: 0.0780 - val_accuracy: 0.9768 - val_precision: 0.9768\n",
            "Epoch 4/10\n",
            "122/122 [==============================] - 1s 6ms/step - loss: 0.0109 - accuracy: 0.9969 - precision: 0.9969 - val_loss: 0.0899 - val_accuracy: 0.9812 - val_precision: 0.9812\n",
            "Epoch 5/10\n",
            "122/122 [==============================] - 1s 5ms/step - loss: 0.0064 - accuracy: 0.9987 - precision: 0.9987 - val_loss: 0.0840 - val_accuracy: 0.9812 - val_precision: 0.9812\n",
            "Epoch 6/10\n",
            "122/122 [==============================] - 1s 5ms/step - loss: 0.0043 - accuracy: 0.9992 - precision: 0.9992 - val_loss: 0.0910 - val_accuracy: 0.9812 - val_precision: 0.9812\n",
            "Epoch 7/10\n",
            "122/122 [==============================] - 1s 6ms/step - loss: 0.0032 - accuracy: 0.9995 - precision: 0.9995 - val_loss: 0.0926 - val_accuracy: 0.9823 - val_precision: 0.9823\n",
            "Epoch 8/10\n",
            "122/122 [==============================] - 1s 5ms/step - loss: 0.0022 - accuracy: 0.9997 - precision: 0.9997 - val_loss: 0.0958 - val_accuracy: 0.9812 - val_precision: 0.9812\n",
            "Epoch 9/10\n",
            "122/122 [==============================] - 1s 5ms/step - loss: 0.0018 - accuracy: 0.9997 - precision: 0.9997 - val_loss: 0.1058 - val_accuracy: 0.9845 - val_precision: 0.9845\n",
            "Epoch 10/10\n",
            "122/122 [==============================] - 1s 5ms/step - loss: 0.0015 - accuracy: 0.9997 - precision: 0.9997 - val_loss: 0.1121 - val_accuracy: 0.9823 - val_precision: 0.9823\n"
          ]
        }
      ],
      "source": [
        "model1 = tuner.hypermodel.build(best_hps)\n",
        "\n",
        "history1 = model1.fit(X_train, np.array(y_train), epochs=10,\n",
        "                    validation_data=(X_dev, np.array(y_dev)))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 20,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 518
        },
        "id": "sft-HFO4fAKY",
        "outputId": "ae456eef-5d2b-43ed-bec7-f8be15308a1f"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "(0.0, 0.28423371384851637)"
            ]
          },
          "execution_count": 20,
          "metadata": {},
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 1152x576 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "plt.figure(figsize=(16, 8))\n",
        "plt.subplot(1, 2, 1)\n",
        "plot_graphs(history1, 'accuracy')\n",
        "plt.ylim(None, 1)\n",
        "plt.subplot(1, 2, 2)\n",
        "plot_graphs(history1, 'loss')\n",
        "plt.ylim(0, None)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 55,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "6vhbPh_WfG8g",
        "outputId": "445e2d00-cd7c-4b37-d677-54793f76ab14"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "         ham       0.99      0.97      0.98       270\n",
            "        spam       0.93      0.98      0.95       118\n",
            "\n",
            "    accuracy                           0.97       388\n",
            "   macro avg       0.96      0.97      0.97       388\n",
            "weighted avg       0.97      0.97      0.97       388\n",
            "\n"
          ]
        }
      ],
      "source": [
        "y_pred = model1.predict(X_test) > 0.5\n",
        "\n",
        "target_names = [\"ham\", \"spam\"]\n",
        "print(classification_report(np.array(y_test), y_pred, target_names=target_names))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bn5c9hwp3PiP"
      },
      "source": [
        "Above we used the **keras_tuner** package to fine-tune the **learning rate** and the **# of hidden units** per layer. We trained the model with the best hyperparameters and got an **f1score** of $0.95$ slightly better from our base model ($0.94$)."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 56,
      "metadata": {
        "id": "7uMFLEXjFQhK"
      },
      "outputs": [],
      "source": [
        "report[\"model_with_best_hp\"] = classification_report(np.array(y_test), y_pred, target_names=target_names, output_dict=True)[\"spam\"][\"f1-score\"]"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "FStNYrJT4Dgh"
      },
      "source": [
        "# Using best hyperparameters - Regularization\n",
        "\n",
        "Next, we are using the best hyperparameters of the above step and build a new model adding some regularization techniques. \n",
        "\n",
        "In [Model 2](https://colab.research.google.com/drive/1uLm9VaTXfUhve3G7E5foxKs5_2ni40ft#scrollTo=WY7Vlfno5mx8&line=1&uniqifier=1) we add two *BatchNormalization*. Previously we mentioned why we use *he_normal* to initialize our inputs' weights. With *BatchNormalization* we actually do the same for the outputs of each layer. Thus, we help our model even more to overpass the vanishing/exploding gradients issue.\n",
        "\n",
        "In [Model 3](https://colab.research.google.com/drive/1uLm9VaTXfUhve3G7E5foxKs5_2ni40ft#scrollTo=22tmnHNA5wma&line=1&uniqifier=1) we add two regularization layers. **Dropout** technique helps our network to reduce overfitting by shutting down randomly a portion of hidden units per layer in each epoch. Thus, the network doesn't lie on some neurons but instead learns weights to use more of them."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WY7Vlfno5mx8"
      },
      "source": [
        "## Model 2 - BatchNormalization"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 22,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "HPoCoEKY4G_A",
        "outputId": "d614d6fc-8e34-4ade-ce6b-2d364eb8f3ea"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Model: \"sequential_2\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " dense_6 (Dense)             (None, 64)                192064    \n",
            "                                                                 \n",
            " batch_normalization (BatchN  (None, 64)               256       \n",
            " ormalization)                                                   \n",
            "                                                                 \n",
            " dense_7 (Dense)             (None, 320)               20800     \n",
            "                                                                 \n",
            " batch_normalization_1 (Batc  (None, 320)              1280      \n",
            " hNormalization)                                                 \n",
            "                                                                 \n",
            " dense_8 (Dense)             (None, 1)                 321       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 214,721\n",
            "Trainable params: 213,953\n",
            "Non-trainable params: 768\n",
            "_________________________________________________________________\n"
          ]
        }
      ],
      "source": [
        "model2 = keras.models.Sequential([\n",
        "    keras.layers.InputLayer(input_shape=(3000,)),\n",
        "    keras.layers.Dense(64, activation=\"relu\", kernel_initializer=\"he_normal\"),\n",
        "    keras.layers.BatchNormalization(),\n",
        "    keras.layers.Dense(320, activation=\"relu\", kernel_initializer=\"he_normal\"),\n",
        "    keras.layers.BatchNormalization(),\n",
        "    keras.layers.Dense(1, activation=\"sigmoid\")                 \n",
        "])\n",
        "\n",
        "model2.summary()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 23,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "t3aZvaxE4rN8",
        "outputId": "49a811e7-2fae-4cdf-a9ef-1375f97ef96d"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1/10\n",
            "122/122 [==============================] - 2s 7ms/step - loss: 0.2435 - accuracy: 0.9028 - recall: 0.9028 - precision: 0.9028 - val_loss: 0.5628 - val_accuracy: 0.7149 - val_recall: 0.7149 - val_precision: 0.7149\n",
            "Epoch 2/10\n",
            "122/122 [==============================] - 1s 5ms/step - loss: 0.0493 - accuracy: 0.9863 - recall: 0.9863 - precision: 0.9863 - val_loss: 0.4355 - val_accuracy: 0.7691 - val_recall: 0.7691 - val_precision: 0.7691\n",
            "Epoch 3/10\n",
            "122/122 [==============================] - 1s 6ms/step - loss: 0.0279 - accuracy: 0.9923 - recall: 0.9923 - precision: 0.9923 - val_loss: 0.2218 - val_accuracy: 0.8807 - val_recall: 0.8807 - val_precision: 0.8807\n",
            "Epoch 4/10\n",
            "122/122 [==============================] - 1s 5ms/step - loss: 0.0143 - accuracy: 0.9961 - recall: 0.9961 - precision: 0.9961 - val_loss: 0.0970 - val_accuracy: 0.9591 - val_recall: 0.9591 - val_precision: 0.9591\n",
            "Epoch 5/10\n",
            "122/122 [==============================] - 1s 5ms/step - loss: 0.0102 - accuracy: 0.9974 - recall: 0.9974 - precision: 0.9974 - val_loss: 0.0740 - val_accuracy: 0.9735 - val_recall: 0.9735 - val_precision: 0.9735\n",
            "Epoch 6/10\n",
            "122/122 [==============================] - 1s 5ms/step - loss: 0.0068 - accuracy: 0.9985 - recall: 0.9985 - precision: 0.9985 - val_loss: 0.0817 - val_accuracy: 0.9790 - val_recall: 0.9790 - val_precision: 0.9790\n",
            "Epoch 7/10\n",
            "122/122 [==============================] - 1s 5ms/step - loss: 0.0075 - accuracy: 0.9979 - recall: 0.9979 - precision: 0.9979 - val_loss: 0.0872 - val_accuracy: 0.9790 - val_recall: 0.9790 - val_precision: 0.9790\n",
            "Epoch 8/10\n",
            "122/122 [==============================] - 1s 6ms/step - loss: 0.0213 - accuracy: 0.9941 - recall: 0.9941 - precision: 0.9941 - val_loss: 0.1210 - val_accuracy: 0.9646 - val_recall: 0.9646 - val_precision: 0.9646\n",
            "Epoch 9/10\n",
            "122/122 [==============================] - 1s 5ms/step - loss: 0.0177 - accuracy: 0.9943 - recall: 0.9943 - precision: 0.9943 - val_loss: 0.0954 - val_accuracy: 0.9669 - val_recall: 0.9669 - val_precision: 0.9669\n",
            "Epoch 10/10\n",
            "122/122 [==============================] - 1s 6ms/step - loss: 0.0070 - accuracy: 0.9979 - recall: 0.9979 - precision: 0.9979 - val_loss: 0.1035 - val_accuracy: 0.9702 - val_recall: 0.9702 - val_precision: 0.9702\n"
          ]
        }
      ],
      "source": [
        "optimizer = keras.optimizers.Adam(learning_rate=0.001)\n",
        "\n",
        "model2.compile(loss=keras.losses.BinaryCrossentropy(),\n",
        "              optimizer=optimizer,\n",
        "              metrics=[\n",
        "                tf.keras.metrics.BinaryAccuracy(name='accuracy'),\n",
        "                tf.keras.metrics.BinaryAccuracy(name='recall'),\n",
        "                tf.keras.metrics.BinaryAccuracy(name='precision')\n",
        "              ])\n",
        "\n",
        "history2 = model2.fit(X_train, np.array(y_train), epochs=10,\n",
        "                    validation_data=(X_dev, np.array(y_dev)))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 24,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 518
        },
        "id": "a5E3rrQx5UfV",
        "outputId": "409cc77d-c5b9-4d59-9c40-df1cac5e88ff"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "(0.0, 0.5905807374510914)"
            ]
          },
          "execution_count": 24,
          "metadata": {},
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 1152x576 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "plt.figure(figsize=(16, 8))\n",
        "plt.subplot(1, 2, 1)\n",
        "plot_graphs(history2, 'accuracy')\n",
        "plt.ylim(None, 1)\n",
        "plt.subplot(1, 2, 2)\n",
        "plot_graphs(history2, 'loss')\n",
        "plt.ylim(0, None)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 57,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "3Ks8LyhRtH_q",
        "outputId": "c81de950-c5b5-4fb2-af86-3518fa055567"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "         ham       0.98      0.96      0.97       270\n",
            "        spam       0.92      0.97      0.94       118\n",
            "\n",
            "    accuracy                           0.96       388\n",
            "   macro avg       0.95      0.96      0.96       388\n",
            "weighted avg       0.96      0.96      0.96       388\n",
            "\n"
          ]
        }
      ],
      "source": [
        "y_pred = model2.predict(X_test) > 0.5\n",
        "\n",
        "target_names = [\"ham\", \"spam\"]\n",
        "print(classification_report(np.array(y_test), y_pred, target_names=target_names))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 58,
      "metadata": {
        "id": "d8C8Ax-kFdKD"
      },
      "outputs": [],
      "source": [
        "report[\"nn_with_bn\"] = classification_report(np.array(y_test), y_pred, target_names=target_names, output_dict=True)[\"spam\"][\"f1-score\"]"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "22tmnHNA5wma"
      },
      "source": [
        "## Model 3 - Dropout"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 26,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "sPL4C8kw__xo",
        "outputId": "397f5333-42a1-4c8d-896b-613a42b1c926"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Model: \"sequential_3\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " dense_9 (Dense)             (None, 64)                192064    \n",
            "                                                                 \n",
            " dropout (Dropout)           (None, 64)                0         \n",
            "                                                                 \n",
            " dense_10 (Dense)            (None, 320)               20800     \n",
            "                                                                 \n",
            " dropout_1 (Dropout)         (None, 320)               0         \n",
            "                                                                 \n",
            " dense_11 (Dense)            (None, 1)                 321       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 213,185\n",
            "Trainable params: 213,185\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "Model: \"sequential_3\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " dense_9 (Dense)             (None, 64)                192064    \n",
            "                                                                 \n",
            " dropout (Dropout)           (None, 64)                0         \n",
            "                                                                 \n",
            " dense_10 (Dense)            (None, 320)               20800     \n",
            "                                                                 \n",
            " dropout_1 (Dropout)         (None, 320)               0         \n",
            "                                                                 \n",
            " dense_11 (Dense)            (None, 1)                 321       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 213,185\n",
            "Trainable params: 213,185\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ],
      "source": [
        "model3 = keras.models.Sequential([\n",
        "    keras.layers.InputLayer(input_shape=(3000,)),\n",
        "    keras.layers.Dense(64, activation=\"relu\", kernel_initializer=\"he_normal\"),\n",
        "    keras.layers.Dropout(0.4),\n",
        "    keras.layers.Dense(320, activation=\"relu\", kernel_initializer=\"he_normal\"),\n",
        "    keras.layers.Dropout(0.4),\n",
        "    keras.layers.Dense(1, activation=\"sigmoid\")                 \n",
        "])\n",
        "\n",
        "model3.summary()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 27,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "7rcceLwuV7L6",
        "outputId": "b346deb4-bf6f-4da8-c4d0-00be58c80be4"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1/10\n",
            "122/122 [==============================] - 1s 7ms/step - loss: 0.3371 - accuracy: 0.8404 - recall: 0.8404 - precision: 0.8404 - val_loss: 0.1033 - val_accuracy: 0.9735 - val_recall: 0.9735 - val_precision: 0.9735\n",
            "Epoch 2/10\n",
            "122/122 [==============================] - 1s 5ms/step - loss: 0.0771 - accuracy: 0.9812 - recall: 0.9812 - precision: 0.9812 - val_loss: 0.0613 - val_accuracy: 0.9790 - val_recall: 0.9790 - val_precision: 0.9790\n",
            "Epoch 3/10\n",
            "122/122 [==============================] - 1s 5ms/step - loss: 0.0343 - accuracy: 0.9899 - recall: 0.9899 - precision: 0.9899 - val_loss: 0.0583 - val_accuracy: 0.9812 - val_recall: 0.9812 - val_precision: 0.9812\n",
            "Epoch 4/10\n",
            "122/122 [==============================] - 1s 5ms/step - loss: 0.0238 - accuracy: 0.9918 - recall: 0.9918 - precision: 0.9918 - val_loss: 0.0843 - val_accuracy: 0.9768 - val_recall: 0.9768 - val_precision: 0.9768\n",
            "Epoch 5/10\n",
            "122/122 [==============================] - 1s 5ms/step - loss: 0.0220 - accuracy: 0.9943 - recall: 0.9943 - precision: 0.9943 - val_loss: 0.0526 - val_accuracy: 0.9779 - val_recall: 0.9779 - val_precision: 0.9779\n",
            "Epoch 6/10\n",
            "122/122 [==============================] - 1s 5ms/step - loss: 0.0125 - accuracy: 0.9969 - recall: 0.9969 - precision: 0.9969 - val_loss: 0.0886 - val_accuracy: 0.9801 - val_recall: 0.9801 - val_precision: 0.9801\n",
            "Epoch 7/10\n",
            "122/122 [==============================] - 1s 5ms/step - loss: 0.0072 - accuracy: 0.9987 - recall: 0.9987 - precision: 0.9987 - val_loss: 0.0532 - val_accuracy: 0.9801 - val_recall: 0.9801 - val_precision: 0.9801\n",
            "Epoch 8/10\n",
            "122/122 [==============================] - 1s 5ms/step - loss: 0.0059 - accuracy: 0.9987 - recall: 0.9987 - precision: 0.9987 - val_loss: 0.0645 - val_accuracy: 0.9812 - val_recall: 0.9812 - val_precision: 0.9812\n",
            "Epoch 9/10\n",
            "122/122 [==============================] - 1s 5ms/step - loss: 0.0247 - accuracy: 0.9941 - recall: 0.9941 - precision: 0.9941 - val_loss: 0.0854 - val_accuracy: 0.9845 - val_recall: 0.9845 - val_precision: 0.9845\n",
            "Epoch 10/10\n",
            "122/122 [==============================] - 1s 5ms/step - loss: 0.0200 - accuracy: 0.9961 - recall: 0.9961 - precision: 0.9961 - val_loss: 0.0673 - val_accuracy: 0.9845 - val_recall: 0.9845 - val_precision: 0.9845\n"
          ]
        }
      ],
      "source": [
        "optimizer = keras.optimizers.Adam(learning_rate=0.001)\n",
        "\n",
        "model3.compile(loss=keras.losses.BinaryCrossentropy(),\n",
        "              optimizer=optimizer,\n",
        "              metrics=[\n",
        "                tf.keras.metrics.BinaryAccuracy(name='accuracy'),\n",
        "                tf.keras.metrics.BinaryAccuracy(name='recall'),\n",
        "                tf.keras.metrics.BinaryAccuracy(name='precision')\n",
        "              ])\n",
        "\n",
        "history3 = model3.fit(X_train, np.array(y_train), epochs=10,\n",
        "                    validation_data=(X_dev, np.array(y_dev)))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 28,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 518
        },
        "id": "FuO3QiMU6KJV",
        "outputId": "e8114314-b175-4554-d028-c8e060af8a4c"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "(0.0, 0.3536314441822469)"
            ]
          },
          "execution_count": 28,
          "metadata": {},
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 1152x576 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "plt.figure(figsize=(16, 8))\n",
        "plt.subplot(1, 2, 1)\n",
        "plot_graphs(history3, 'accuracy')\n",
        "plt.ylim(None, 1)\n",
        "plt.subplot(1, 2, 2)\n",
        "plot_graphs(history3, 'loss')\n",
        "plt.ylim(0, None)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 59,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "MqsRBGcwtK_3",
        "outputId": "a2e10287-72cc-4840-e2e5-69d942e0f6c6"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "         ham       0.99      0.98      0.99       270\n",
            "        spam       0.96      0.97      0.97       118\n",
            "\n",
            "    accuracy                           0.98       388\n",
            "   macro avg       0.97      0.98      0.98       388\n",
            "weighted avg       0.98      0.98      0.98       388\n",
            "\n"
          ]
        }
      ],
      "source": [
        "y_pred = model3.predict(X_test) > 0.5\n",
        "\n",
        "target_names = [\"ham\", \"spam\"]\n",
        "print(classification_report(np.array(y_test), y_pred, target_names=target_names))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "J7h0tZ-P6U8Y"
      },
      "source": [
        "As expected *BatchNormalization* smooths the loss function and *Dropout* does better on **f1score** for **Spam** emails."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 60,
      "metadata": {
        "id": "TarXDXebFk3s"
      },
      "outputs": [],
      "source": [
        "report[\"nn_with_dropout\"] = classification_report(np.array(y_test), y_pred, target_names=target_names, output_dict=True)[\"spam\"][\"f1-score\"]"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "PNzTwggJ_niP"
      },
      "source": [
        "## All in one\n",
        "\n",
        "The last model will be the combination of all the above. Thus, we will add *BatchNormalization* and *Dropout* layers."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 30,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "-KevMK1HHCNm",
        "outputId": "20197c2e-ce53-4d10-8a7a-af661204925c"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Model: \"sequential_4\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " dense_12 (Dense)            (None, 64)                192064    \n",
            "                                                                 \n",
            " dropout_2 (Dropout)         (None, 64)                0         \n",
            "                                                                 \n",
            " batch_normalization_2 (Batc  (None, 64)               256       \n",
            " hNormalization)                                                 \n",
            "                                                                 \n",
            " dense_13 (Dense)            (None, 320)               20800     \n",
            "                                                                 \n",
            " dropout_3 (Dropout)         (None, 320)               0         \n",
            "                                                                 \n",
            " batch_normalization_3 (Batc  (None, 320)              1280      \n",
            " hNormalization)                                                 \n",
            "                                                                 \n",
            " dense_14 (Dense)            (None, 1)                 321       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 214,721\n",
            "Trainable params: 213,953\n",
            "Non-trainable params: 768\n",
            "_________________________________________________________________\n"
          ]
        }
      ],
      "source": [
        "model4 = keras.models.Sequential([\n",
        "    keras.layers.InputLayer(input_shape=(3000,)),\n",
        "    keras.layers.Dense(64, activation=\"relu\", kernel_initializer=\"he_normal\"),\n",
        "    keras.layers.Dropout(0.4),\n",
        "    keras.layers.BatchNormalization(),\n",
        "    keras.layers.Dense(320, activation=\"relu\", kernel_initializer=\"he_normal\"),\n",
        "    keras.layers.Dropout(0.4),\n",
        "    keras.layers.BatchNormalization(),\n",
        "    keras.layers.Dense(1, activation=\"sigmoid\")                 \n",
        "])\n",
        "\n",
        "model4.summary()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "lY6D6MaXHXLP"
      },
      "outputs": [],
      "source": [
        "optimizer = keras.optimizers.Adam(learning_rate=0.001)\n",
        "\n",
        "model4.compile(loss=keras.losses.BinaryCrossentropy(),\n",
        "              optimizer=optimizer,\n",
        "              metrics=[\n",
        "                tf.keras.metrics.BinaryAccuracy(name='accuracy'),\n",
        "                tf.keras.metrics.BinaryAccuracy(name='recall'),\n",
        "                tf.keras.metrics.BinaryAccuracy(name='precision')\n",
        "              ])\n",
        "\n",
        "history4 = model4.fit(X_train, np.array(y_train), epochs=10,\n",
        "                    validation_data=(X_dev, np.array(y_dev)))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 36,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 518
        },
        "id": "6pUw5gp-A4VM",
        "outputId": "d13b09ea-2d88-4a1d-ad98-e852f941b101"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "(0.0, 0.47759564509615304)"
            ]
          },
          "execution_count": 36,
          "metadata": {},
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 1152x576 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "plt.figure(figsize=(16, 8))\n",
        "plt.subplot(1, 2, 1)\n",
        "plot_graphs(history4, 'accuracy')\n",
        "plt.ylim(None, 1)\n",
        "plt.subplot(1, 2, 2)\n",
        "plot_graphs(history4, 'loss')\n",
        "plt.ylim(0, None)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 61,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "DKr8ZRM_tNDL",
        "outputId": "037bc348-a170-4215-c907-d5842a390415"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "         ham       0.98      0.99      0.98       270\n",
            "        spam       0.97      0.96      0.96       118\n",
            "\n",
            "    accuracy                           0.98       388\n",
            "   macro avg       0.97      0.97      0.97       388\n",
            "weighted avg       0.98      0.98      0.98       388\n",
            "\n"
          ]
        }
      ],
      "source": [
        "y_pred = model4.predict(X_test) > 0.5\n",
        "\n",
        "target_names = [\"ham\", \"spam\"]\n",
        "print(classification_report(np.array(y_test), y_pred, target_names=target_names))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 62,
      "metadata": {
        "id": "QxYEq-SfFrfy"
      },
      "outputs": [],
      "source": [
        "report[\"nn_with_bn_dropout\"] = classification_report(np.array(y_test), y_pred, target_names=target_names, output_dict=True)[\"spam\"][\"f1-score\"]"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6jrVFPdbBfPw"
      },
      "source": [
        "### Adding BatchNormalization Before Dropout"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 34,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_rDLqHNjBlTc",
        "outputId": "cc8780dc-f499-4a28-cb14-add0d845de1e"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Model: \"sequential_5\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " dense_15 (Dense)            (None, 64)                192064    \n",
            "                                                                 \n",
            " batch_normalization_4 (Batc  (None, 64)               256       \n",
            " hNormalization)                                                 \n",
            "                                                                 \n",
            " dropout_4 (Dropout)         (None, 64)                0         \n",
            "                                                                 \n",
            " dense_16 (Dense)            (None, 320)               20800     \n",
            "                                                                 \n",
            " batch_normalization_5 (Batc  (None, 320)              1280      \n",
            " hNormalization)                                                 \n",
            "                                                                 \n",
            " dropout_5 (Dropout)         (None, 320)               0         \n",
            "                                                                 \n",
            " dense_17 (Dense)            (None, 1)                 321       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 214,721\n",
            "Trainable params: 213,953\n",
            "Non-trainable params: 768\n",
            "_________________________________________________________________\n"
          ]
        }
      ],
      "source": [
        "model5 = keras.models.Sequential([\n",
        "    keras.layers.InputLayer(input_shape=(3000,)),\n",
        "    keras.layers.Dense(64, activation=\"relu\", kernel_initializer=\"he_normal\"),\n",
        "    keras.layers.BatchNormalization(),\n",
        "    keras.layers.Dropout(0.4),\n",
        "    keras.layers.Dense(320, activation=\"relu\", kernel_initializer=\"he_normal\"),\n",
        "    keras.layers.BatchNormalization(),\n",
        "    keras.layers.Dropout(0.4),\n",
        "    keras.layers.Dense(1, activation=\"sigmoid\")                 \n",
        "])\n",
        "\n",
        "model5.summary()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "7IP1oaVKB3l7"
      },
      "outputs": [],
      "source": [
        "optimizer = keras.optimizers.Adam(learning_rate=0.001)\n",
        "\n",
        "model5.compile(loss=keras.losses.BinaryCrossentropy(),\n",
        "              optimizer=optimizer,\n",
        "              metrics=[\n",
        "                tf.keras.metrics.BinaryAccuracy(name='accuracy'),\n",
        "                tf.keras.metrics.BinaryAccuracy(name='recall'),\n",
        "                tf.keras.metrics.BinaryAccuracy(name='precision')\n",
        "              ])\n",
        "\n",
        "history5 = model5.fit(X_train, np.array(y_train), epochs=10,\n",
        "                    validation_data=(X_dev, np.array(y_dev)))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 37,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 518
        },
        "id": "_WECIWi2B-kA",
        "outputId": "f3d4ab46-d4a4-49db-8d8d-a2ece11d484b"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "(0.0, 0.4987489405088127)"
            ]
          },
          "execution_count": 37,
          "metadata": {},
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 1152x576 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "plt.figure(figsize=(16, 8))\n",
        "plt.subplot(1, 2, 1)\n",
        "plot_graphs(history5, 'accuracy')\n",
        "plt.ylim(None, 1)\n",
        "plt.subplot(1, 2, 2)\n",
        "plot_graphs(history5, 'loss')\n",
        "plt.ylim(0, None)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 63,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "jfIG5gNbCI6I",
        "outputId": "aefb0de8-386a-49d0-d62e-3ce636c35fa4"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "         ham       0.99      0.97      0.98       270\n",
            "        spam       0.93      0.97      0.95       118\n",
            "\n",
            "    accuracy                           0.97       388\n",
            "   macro avg       0.96      0.97      0.96       388\n",
            "weighted avg       0.97      0.97      0.97       388\n",
            "\n"
          ]
        }
      ],
      "source": [
        "y_pred = model5.predict(X_test) > 0.5\n",
        "\n",
        "target_names = [\"ham\", \"spam\"]\n",
        "print(classification_report(np.array(y_test), y_pred, target_names=target_names))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 64,
      "metadata": {
        "id": "mwx90dmwF1SX"
      },
      "outputs": [],
      "source": [
        "report[\"nn_bn_first\"] = classification_report(np.array(y_test), y_pred, target_names=target_names, output_dict=True)[\"spam\"][\"f1-score\"]"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VPRLT0-YCS-D"
      },
      "source": [
        "### Adding BatchNormalization Before Activation Function\n",
        "\n",
        "There is a controversy whether the BatchNormalization should be conducted before or after the activation function. In this step, we apply BatchNormalization before the activation function. The performance of this network is worst than the previous. This means that this isn't the appropriate approach for this project."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 39,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "1qipE_77CYaN",
        "outputId": "56b5287a-b4cc-4f4f-de5e-d67227568908"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Model: \"sequential_6\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " dense_18 (Dense)            (None, 64)                192064    \n",
            "                                                                 \n",
            " batch_normalization_6 (Batc  (None, 64)               256       \n",
            " hNormalization)                                                 \n",
            "                                                                 \n",
            " activation (Activation)     (None, 64)                0         \n",
            "                                                                 \n",
            " dropout_6 (Dropout)         (None, 64)                0         \n",
            "                                                                 \n",
            " dense_19 (Dense)            (None, 320)               20800     \n",
            "                                                                 \n",
            " batch_normalization_7 (Batc  (None, 320)              1280      \n",
            " hNormalization)                                                 \n",
            "                                                                 \n",
            " activation_1 (Activation)   (None, 320)               0         \n",
            "                                                                 \n",
            " dropout_7 (Dropout)         (None, 320)               0         \n",
            "                                                                 \n",
            " dense_20 (Dense)            (None, 1)                 321       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 214,721\n",
            "Trainable params: 213,953\n",
            "Non-trainable params: 768\n",
            "_________________________________________________________________\n"
          ]
        }
      ],
      "source": [
        "model6 = keras.models.Sequential([\n",
        "    keras.layers.InputLayer(input_shape=(3000,)),\n",
        "    keras.layers.Dense(64, kernel_initializer=\"he_normal\"),\n",
        "    keras.layers.BatchNormalization(),\n",
        "    keras.layers.Activation(\"relu\"),\n",
        "    keras.layers.Dropout(0.4),\n",
        "    keras.layers.Dense(320, kernel_initializer=\"he_normal\"),\n",
        "    keras.layers.BatchNormalization(),\n",
        "    keras.layers.Activation(\"relu\"),\n",
        "    keras.layers.Dropout(0.4),\n",
        "    keras.layers.Dense(1, activation=\"sigmoid\")                 \n",
        "])\n",
        "\n",
        "model6.summary()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "DtZNi63eCo-3"
      },
      "outputs": [],
      "source": [
        "optimizer = keras.optimizers.Adam(learning_rate=0.001)\n",
        "\n",
        "model6.compile(loss=keras.losses.BinaryCrossentropy(),\n",
        "              optimizer=optimizer,\n",
        "              metrics=[\n",
        "                tf.keras.metrics.BinaryAccuracy(name='accuracy'),\n",
        "                tf.keras.metrics.BinaryAccuracy(name='recall'),\n",
        "                tf.keras.metrics.BinaryAccuracy(name='precision')\n",
        "              ])\n",
        "\n",
        "history6 = model6.fit(X_train, np.array(y_train), epochs=10,\n",
        "                    validation_data=(X_dev, np.array(y_dev)))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 41,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 518
        },
        "id": "NShS4Ks3CuWL",
        "outputId": "832796b8-48af-4df5-d948-3b798355f458"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "(0.0, 0.3713682655245066)"
            ]
          },
          "execution_count": 41,
          "metadata": {},
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 1152x576 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "plt.figure(figsize=(16, 8))\n",
        "plt.subplot(1, 2, 1)\n",
        "plot_graphs(history6, 'accuracy')\n",
        "plt.ylim(None, 1)\n",
        "plt.subplot(1, 2, 2)\n",
        "plot_graphs(history6, 'loss')\n",
        "plt.ylim(0, None)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 65,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "YXBfUf50C0Yj",
        "outputId": "347922ac-c554-459c-e275-33a9d5605c56"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "         ham       1.00      0.93      0.96       270\n",
            "        spam       0.85      0.99      0.92       118\n",
            "\n",
            "    accuracy                           0.95       388\n",
            "   macro avg       0.93      0.96      0.94       388\n",
            "weighted avg       0.95      0.95      0.95       388\n",
            "\n"
          ]
        }
      ],
      "source": [
        "y_pred = model6.predict(X_test) > 0.5\n",
        "\n",
        "target_names = [\"ham\", \"spam\"]\n",
        "print(classification_report(np.array(y_test), y_pred, target_names=target_names))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 66,
      "metadata": {
        "id": "ueoH2l9eGE2L"
      },
      "outputs": [],
      "source": [
        "report[\"nn_activation_last\"] = classification_report(np.array(y_test), y_pred, target_names=target_names, output_dict=True)[\"spam\"][\"f1-score\"]"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JYropmLLOlfH"
      },
      "source": [
        "\n",
        "\n",
        "---\n",
        "\n",
        "\n",
        "---\n",
        "\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 69,
      "metadata": {
        "id": "z47JW_PGGOm-"
      },
      "outputs": [],
      "source": [
        "report_df = pd.DataFrame.from_dict(report, orient=\"index\", columns=[\"f1-score\"], dtype=np.float16)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 91,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 544
        },
        "id": "alBn8zKxGREE",
        "outputId": "e569f4dc-ced5-4bce-8c16-9d36343650e5"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "plt.rcdefaults()\n",
        "fig, ax = plt.subplots()\n",
        "\n",
        "ax.barh(y=np.arange(len(report.keys())), width=report_df[\"f1-score\"], color=\"gray\")\n",
        "ax.barh(y=3, width=report_df[\"f1-score\"].max(), color=\"red\")\n",
        "ax.set_yticks(ticks=np.arange(len(report.keys())))\n",
        "ax.set_yticklabels(labels=[\"Base NN\", \"NN /w Best HP\", \"NN /w BatcNormalization (BN)\", \"NN /w Dropout\", \"NN: Dropout->BN\", \"NN: BN->Dropout\", \"NN: BN->Activation->Dropout\"], rotation=45)\n",
        "ax.invert_yaxis()\n",
        "ax.set_xlabel(\"F1-score on SPAM class\")\n",
        "ax.set_title(\"NN-architectures' Performance\")\n",
        "\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "quBzE2TQQZQB"
      },
      "source": [
        "# Save Model"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 92,
      "metadata": {
        "id": "46hfsvebQhGQ"
      },
      "outputs": [],
      "source": [
        "model3.save(\"email_classifier.h5\")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yup7MiVuMpUe"
      },
      "source": [
        "# Conclusions\n",
        "\n",
        "In this project, we built a *SPAM classifier*. Our dataset consists of 5000+ instances with 3000 features each (we don't include the id and the prediction). \n",
        "\n",
        "We split the dataset into three sets a **train**, a **develop**, and a **test** set. The development set is used for the validation and the test for the final evaluation on **f1-score**. We kept the model with the best f1-score on the SPAM class on the test set.\n",
        "\n",
        "We built a base neural network randomly chosen to have 2 hidden layers and the output. Then, we fine-tuned it to get the best **hyperparameters** (learning rate and a number of hidden units per layer). Moreover, we add a regularization technique to reduce **variance**.\n",
        "\n",
        "**Batchnormalization** didn't add much value to the network possibly because we have a small dataset but as expected it smoothed the loss curve. Dropout on the other hand added value to the network and increased the performance by about 3% on the f1-score.\n",
        "\n",
        "We kept for the final submission the neural network with the dropout layers only based."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "i7K6_Nmiqn8j"
      },
      "source": [
        "### References\n",
        "\n",
        "\n",
        "\n",
        "*   [Keras Tuner](https://www.tensorflow.org/tutorials/keras/keras_tuner)\n",
        "*   [Hands-On Machine Learning with Scikit-Learn, Keras & Tensorflow](https://github.com/ageron/handson-ml2)\n",
        "\n"
      ]
    }
  ],
  "metadata": {
    "colab": {
      "collapsed_sections": [
        "9Xg67jjx1QH0"
      ],
      "name": "email classification.ipynb",
      "provenance": []
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}