ANN_Customer_Churn_Model.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "source": [
        "# Introduction\n",
        "\n",
        "This project is for a bank to detect it's customer churn. The bank has seen that there was a lot of churn over the last 6 months and it wants us to create a model that can predict whether, given certain specific details about a customer, would that customer stay or leave the bank.\n",
        "\n",
        "This model has been created using Artificial Neural Network to classify whether a customer would stay or leave the bank.\n",
        "\n",
        "A value of \"True\" means that the customer would leave the bank.\n",
        "A value of \"False\" means that the customer would stay with the bank."
      ],
      "metadata": {
        "id": "G7FVgZo-QC_m"
      }
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lP6JLo1tGNBg"
      },
      "source": [
        "# Artificial Neural Network"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gWZyYmS_UE_L"
      },
      "source": [
        "### Importing the libraries"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "import tensorflow as tf"
      ],
      "metadata": {
        "id": "Td-DvcAdSDw5"
      },
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "tf.__version__  # checking the version of TensorFlow"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "id": "OMUY43qSTU0K",
        "outputId": "f6b0c584-f2f7-42ce-9829-31dde597cf4f"
      },
      "execution_count": 41,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'2.15.0'"
            ],
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            }
          },
          "metadata": {},
          "execution_count": 41
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "t3HC4IJJSXPj",
        "outputId": "820d309f-0960-428f-8738-ce6cef28830f"
      },
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/drive\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1E0Q3aoKUCRX"
      },
      "source": [
        "## Part 1 - Data Preprocessing"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cKWAkFVGUU0Z"
      },
      "source": [
        "### Importing the dataset"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Creating the X and y data arrays after reading the dataset\n",
        "\n",
        "dataset = pd.read_csv('/content/drive/MyDrive/Churn_Modelling.csv')\n",
        "X = dataset.iloc[:, 3:-1].values\n",
        "y = dataset.iloc[:, -1].values"
      ],
      "metadata": {
        "id": "PMRFrRvWTx4l"
      },
      "execution_count": 42,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "print(X)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_bqIHVE9UtMH",
        "outputId": "2bab3446-1f55-4742-ad37-956de9653f41"
      },
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[[619 'France' 'Female' ... 1 1 101348.88]\n",
            " [608 'Spain' 'Female' ... 0 1 112542.58]\n",
            " [502 'France' 'Female' ... 1 0 113931.57]\n",
            " ...\n",
            " [709 'France' 'Female' ... 0 1 42085.58]\n",
            " [772 'Germany' 'Male' ... 1 0 92888.52]\n",
            " [792 'France' 'Female' ... 1 0 38190.78]]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(y)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "v0nEMV8AUt5L",
        "outputId": "0ee14feb-b138-4e26-8e65-567f5a9b3784"
      },
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[1 0 1 ... 1 1 0]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "N6bQ0UgSU-NJ"
      },
      "source": [
        "### Encoding categorical data"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "le5MJreAbW52"
      },
      "source": [
        "Label Encoding the \"Gender\" column"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.preprocessing import LabelEncoder\n",
        "le = LabelEncoder()\n",
        "X[:, 2]= le.fit_transform(X[:, 2])"
      ],
      "metadata": {
        "id": "88fdALe0VGJD"
      },
      "execution_count": 8,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "print(X)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "oZTMKgsgVsa_",
        "outputId": "2cdc748e-8b50-48c2-c024-198e05e94c04"
      },
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[[619 'France' 0 ... 1 1 101348.88]\n",
            " [608 'Spain' 0 ... 0 1 112542.58]\n",
            " [502 'France' 0 ... 1 0 113931.57]\n",
            " ...\n",
            " [709 'France' 0 ... 0 1 42085.58]\n",
            " [772 'Germany' 1 ... 1 0 92888.52]\n",
            " [792 'France' 0 ... 1 0 38190.78]]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CUxGZezpbMcb"
      },
      "source": [
        "One Hot Encoding the \"Geography\" column"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.preprocessing import OneHotEncoder\n",
        "from sklearn.compose import ColumnTransformer\n",
        "ct = ColumnTransformer(transformers=[(\"encoder\", OneHotEncoder(), [1])], remainder=\"passthrough\")\n",
        "X = np.array(ct.fit_transform(X))"
      ],
      "metadata": {
        "id": "NlR7mB6jVyG8"
      },
      "execution_count": 40,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "print(X)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "GXEjSDFBWOAN",
        "outputId": "670855c8-9d6e-441e-d8ea-e496205613dc"
      },
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[[1.0 0.0 0.0 ... 1 1 101348.88]\n",
            " [0.0 0.0 1.0 ... 0 1 112542.58]\n",
            " [1.0 0.0 0.0 ... 1 0 113931.57]\n",
            " ...\n",
            " [1.0 0.0 0.0 ... 0 1 42085.58]\n",
            " [0.0 1.0 0.0 ... 1 0 92888.52]\n",
            " [1.0 0.0 0.0 ... 1 0 38190.78]]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vHol938cW8zd"
      },
      "source": [
        "### Splitting the dataset into the Training set and Test set"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.2, random_state = 0)"
      ],
      "metadata": {
        "id": "Jj-V-wJKWccJ"
      },
      "execution_count": 12,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "RE_FcHyfV3TQ"
      },
      "source": [
        "### Feature Scaling"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# We will use StandardScaler for standardizing everything in X_train and X_test\n",
        "\n",
        "from sklearn.preprocessing import StandardScaler\n",
        "sc = StandardScaler()\n",
        "X_train = sc.fit_transform(X_train)\n",
        "X_test = sc.transform(X_test)"
      ],
      "metadata": {
        "id": "AoR6kNXVWmJv"
      },
      "execution_count": 13,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-zfEzkRVXIwF"
      },
      "source": [
        "## Part 2 - Building the ANN"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "KvdeScabXtlB"
      },
      "source": [
        "### Initializing the ANN"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# We will create a variable for the ANN as an object of the Sequential class\n",
        "\n",
        "ann = tf.keras.models.Sequential()"
      ],
      "metadata": {
        "id": "cyqI7PWyXRxt"
      },
      "execution_count": 14,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rP6urV6SX7kS"
      },
      "source": [
        "### Adding the input layer and the first hidden layer"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Using rectifier as the activation function here\n",
        "\n",
        "ann.add(tf.keras.layers.Dense(units=6, activation=\"relu\"))"
      ],
      "metadata": {
        "id": "577KhzIaYCHU"
      },
      "execution_count": 19,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BELWAc_8YJze"
      },
      "source": [
        "### Adding the second hidden layer"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Using rectifier as the activation again\n",
        "\n",
        "ann.add(tf.keras.layers.Dense(units=6, activation=\"relu\"))"
      ],
      "metadata": {
        "id": "MqWDymyMGYbM"
      },
      "execution_count": 18,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OyNEe6RXYcU4"
      },
      "source": [
        "### Adding the output layer"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Final output layer will have sigmoid activation (binary) and unit would be 1 as we want only 1 result set as \"exited\"\n",
        "# if it was categorical classification we would've used softmax activation\n",
        "\n",
        "ann.add(tf.keras.layers.Dense(units=1, activation=\"sigmoid\"))"
      ],
      "metadata": {
        "id": "c6BE_FKhGbZQ"
      },
      "execution_count": 33,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JT4u2S1_Y4WG"
      },
      "source": [
        "## Part 3 - Training the ANN"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8GWlJChhY_ZI"
      },
      "source": [
        "### Compiling the ANN"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# We will use Stochastic Gradient Descent (adam) optimizer\n",
        "# We will use binary_crossentropy by default as it is binary classification. For categorical we would've used categorical_crossentropy\n",
        "\n",
        "ann.compile(optimizer=\"adam\", loss=\"binary_crossentropy\", metrics=[\"accuracy\"])"
      ],
      "metadata": {
        "id": "MIi3e35JHrJb"
      },
      "execution_count": 34,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0QR_G5u7ZLSM"
      },
      "source": [
        "### Training the ANN on the Training set"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "ann.fit(X_train, y_train, batch_size=32, epochs=100)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ropCAJIPJAg5",
        "outputId": "0d33e090-b2d6-4b57-9803-064ae0cccca9"
      },
      "execution_count": 35,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/100\n",
            "250/250 [==============================] - 2s 3ms/step - loss: 0.6574 - accuracy: 0.8309\n",
            "Epoch 2/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.5972 - accuracy: 0.8080\n",
            "Epoch 3/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.5506 - accuracy: 0.7960\n",
            "Epoch 4/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.5143 - accuracy: 0.7960\n",
            "Epoch 5/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.4860 - accuracy: 0.7960\n",
            "Epoch 6/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.4627 - accuracy: 0.8185\n",
            "Epoch 7/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.4456 - accuracy: 0.8646\n",
            "Epoch 8/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.4304 - accuracy: 0.8662\n",
            "Epoch 9/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.4196 - accuracy: 0.8633\n",
            "Epoch 10/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.4082 - accuracy: 0.8652\n",
            "Epoch 11/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.4006 - accuracy: 0.8665\n",
            "Epoch 12/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3940 - accuracy: 0.8641\n",
            "Epoch 13/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3878 - accuracy: 0.8651\n",
            "Epoch 14/100\n",
            "250/250 [==============================] - 1s 4ms/step - loss: 0.3831 - accuracy: 0.8645\n",
            "Epoch 15/100\n",
            "250/250 [==============================] - 1s 4ms/step - loss: 0.3786 - accuracy: 0.8652\n",
            "Epoch 16/100\n",
            "250/250 [==============================] - 1s 4ms/step - loss: 0.3754 - accuracy: 0.8648\n",
            "Epoch 17/100\n",
            "250/250 [==============================] - 1s 4ms/step - loss: 0.3726 - accuracy: 0.8640\n",
            "Epoch 18/100\n",
            "250/250 [==============================] - 1s 4ms/step - loss: 0.3695 - accuracy: 0.8639\n",
            "Epoch 19/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3675 - accuracy: 0.8640\n",
            "Epoch 20/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3648 - accuracy: 0.8649\n",
            "Epoch 21/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3620 - accuracy: 0.8645\n",
            "Epoch 22/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3600 - accuracy: 0.8656\n",
            "Epoch 23/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3587 - accuracy: 0.8676\n",
            "Epoch 24/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3567 - accuracy: 0.8660\n",
            "Epoch 25/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3558 - accuracy: 0.8652\n",
            "Epoch 26/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3541 - accuracy: 0.8644\n",
            "Epoch 27/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3537 - accuracy: 0.8669\n",
            "Epoch 28/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3525 - accuracy: 0.8652\n",
            "Epoch 29/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3513 - accuracy: 0.8661\n",
            "Epoch 30/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3509 - accuracy: 0.8665\n",
            "Epoch 31/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3498 - accuracy: 0.8674\n",
            "Epoch 32/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3489 - accuracy: 0.8679\n",
            "Epoch 33/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3475 - accuracy: 0.8686\n",
            "Epoch 34/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3474 - accuracy: 0.8665\n",
            "Epoch 35/100\n",
            "250/250 [==============================] - 1s 4ms/step - loss: 0.3475 - accuracy: 0.8662\n",
            "Epoch 36/100\n",
            "250/250 [==============================] - 1s 4ms/step - loss: 0.3460 - accuracy: 0.8677\n",
            "Epoch 37/100\n",
            "250/250 [==============================] - 1s 4ms/step - loss: 0.3453 - accuracy: 0.8673\n",
            "Epoch 38/100\n",
            "250/250 [==============================] - 1s 5ms/step - loss: 0.3453 - accuracy: 0.8671\n",
            "Epoch 39/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3445 - accuracy: 0.8673\n",
            "Epoch 40/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3449 - accuracy: 0.8671\n",
            "Epoch 41/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3439 - accuracy: 0.8662\n",
            "Epoch 42/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3428 - accuracy: 0.8674\n",
            "Epoch 43/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3430 - accuracy: 0.8692\n",
            "Epoch 44/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3426 - accuracy: 0.8695\n",
            "Epoch 45/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3420 - accuracy: 0.8679\n",
            "Epoch 46/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3426 - accuracy: 0.8675\n",
            "Epoch 47/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3416 - accuracy: 0.8685\n",
            "Epoch 48/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3413 - accuracy: 0.8691\n",
            "Epoch 49/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3404 - accuracy: 0.8666\n",
            "Epoch 50/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3407 - accuracy: 0.8694\n",
            "Epoch 51/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3403 - accuracy: 0.8666\n",
            "Epoch 52/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3402 - accuracy: 0.8671\n",
            "Epoch 53/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3391 - accuracy: 0.8689\n",
            "Epoch 54/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3399 - accuracy: 0.8666\n",
            "Epoch 55/100\n",
            "250/250 [==============================] - 1s 4ms/step - loss: 0.3402 - accuracy: 0.8658\n",
            "Epoch 56/100\n",
            "250/250 [==============================] - 1s 4ms/step - loss: 0.3402 - accuracy: 0.8655\n",
            "Epoch 57/100\n",
            "250/250 [==============================] - 1s 4ms/step - loss: 0.3388 - accuracy: 0.8674\n",
            "Epoch 58/100\n",
            "250/250 [==============================] - 1s 4ms/step - loss: 0.3381 - accuracy: 0.8690\n",
            "Epoch 59/100\n",
            "250/250 [==============================] - 1s 4ms/step - loss: 0.3391 - accuracy: 0.8681\n",
            "Epoch 60/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3383 - accuracy: 0.8679\n",
            "Epoch 61/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3388 - accuracy: 0.8665\n",
            "Epoch 62/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3382 - accuracy: 0.8687\n",
            "Epoch 63/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3381 - accuracy: 0.8680\n",
            "Epoch 64/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3376 - accuracy: 0.8676\n",
            "Epoch 65/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3377 - accuracy: 0.8685\n",
            "Epoch 66/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3378 - accuracy: 0.8666\n",
            "Epoch 67/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3381 - accuracy: 0.8675\n",
            "Epoch 68/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3370 - accuracy: 0.8691\n",
            "Epoch 69/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3368 - accuracy: 0.8670\n",
            "Epoch 70/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3367 - accuracy: 0.8686\n",
            "Epoch 71/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3368 - accuracy: 0.8674\n",
            "Epoch 72/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3367 - accuracy: 0.8684\n",
            "Epoch 73/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3370 - accuracy: 0.8669\n",
            "Epoch 74/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3364 - accuracy: 0.8689\n",
            "Epoch 75/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3360 - accuracy: 0.8675\n",
            "Epoch 76/100\n",
            "250/250 [==============================] - 1s 4ms/step - loss: 0.3361 - accuracy: 0.8675\n",
            "Epoch 77/100\n",
            "250/250 [==============================] - 1s 4ms/step - loss: 0.3367 - accuracy: 0.8648\n",
            "Epoch 78/100\n",
            "250/250 [==============================] - 1s 4ms/step - loss: 0.3356 - accuracy: 0.8680\n",
            "Epoch 79/100\n",
            "250/250 [==============================] - 1s 4ms/step - loss: 0.3367 - accuracy: 0.8674\n",
            "Epoch 80/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3359 - accuracy: 0.8687\n",
            "Epoch 81/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3353 - accuracy: 0.8680\n",
            "Epoch 82/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3351 - accuracy: 0.8687\n",
            "Epoch 83/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3357 - accuracy: 0.8683\n",
            "Epoch 84/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3354 - accuracy: 0.8664\n",
            "Epoch 85/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3352 - accuracy: 0.8674\n",
            "Epoch 86/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3352 - accuracy: 0.8680\n",
            "Epoch 87/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3355 - accuracy: 0.8671\n",
            "Epoch 88/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3348 - accuracy: 0.8686\n",
            "Epoch 89/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3350 - accuracy: 0.8664\n",
            "Epoch 90/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3342 - accuracy: 0.8691\n",
            "Epoch 91/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3345 - accuracy: 0.8681\n",
            "Epoch 92/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3340 - accuracy: 0.8676\n",
            "Epoch 93/100\n",
            "250/250 [==============================] - 1s 2ms/step - loss: 0.3338 - accuracy: 0.8680\n",
            "Epoch 94/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3339 - accuracy: 0.8698\n",
            "Epoch 95/100\n",
            "250/250 [==============================] - 1s 3ms/step - loss: 0.3336 - accuracy: 0.8681\n",
            "Epoch 96/100\n",
            "250/250 [==============================] - 1s 4ms/step - loss: 0.3344 - accuracy: 0.8695\n",
            "Epoch 97/100\n",
            "250/250 [==============================] - 1s 4ms/step - loss: 0.3339 - accuracy: 0.8674\n",
            "Epoch 98/100\n",
            "250/250 [==============================] - 1s 4ms/step - loss: 0.3337 - accuracy: 0.8671\n",
            "Epoch 99/100\n",
            "250/250 [==============================] - 1s 4ms/step - loss: 0.3338 - accuracy: 0.8680\n",
            "Epoch 100/100\n",
            "250/250 [==============================] - 1s 4ms/step - loss: 0.3338 - accuracy: 0.8687\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.src.callbacks.History at 0x7d8a9bf1a740>"
            ]
          },
          "metadata": {},
          "execution_count": 35
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tJj5k2MxZga3"
      },
      "source": [
        "## Part 4 - Making the predictions and evaluating the model"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "84QFoqGYeXHL"
      },
      "source": [
        "### Predicting the result of a single observation"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CGRo3eacgDdC"
      },
      "source": [
        "**Test**\n",
        "\n",
        "Use the ANN model to predict if the customer with the following informations will leave the bank:\n",
        "\n",
        "Geography: France\n",
        "\n",
        "Credit Score: 600\n",
        "\n",
        "Gender: Male\n",
        "\n",
        "Age: 40 years old\n",
        "\n",
        "Tenure: 3 years\n",
        "\n",
        "Balance: \\$ 60000\n",
        "\n",
        "Number of Products: 2\n",
        "\n",
        "Does this customer have a credit card ? Yes\n",
        "\n",
        "Is this customer an Active Member: Yes\n",
        "\n",
        "Estimated Salary: \\$ 50000\n",
        "\n",
        "So, should we say goodbye to that customer ?"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZhU1LTgPg-kH"
      },
      "source": [
        "**Solution**"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(ann.predict(sc.transform([[1, 0, 0, 600, 1, 40, 3, 60000, 2, 1, 1, 50000]])) > 0.5)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "W9IbcKYIK3kW",
        "outputId": "548d7dcc-8bc4-4f39-8eff-936d55444d43"
      },
      "execution_count": 36,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "1/1 [==============================] - 0s 134ms/step\n",
            "[[False]]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wGjx94g2n7OV"
      },
      "source": [
        "Therefore, our ANN model predicts that this customer stays in the bank!\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "u7yx47jPZt11"
      },
      "source": [
        "### Predicting the Test set results"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "y_pred = ann.predict(X_test)\n",
        "y_pred = (y_pred > 0.5)\n",
        "print(np.concatenate((y_pred.reshape(len(y_pred),1), y_test.reshape(len(y_test),1)),1))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "vkv6Yt6tNIQ5",
        "outputId": "d6dcc991-998d-43b0-f1f7-e92cbcd1f0e4"
      },
      "execution_count": 37,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "63/63 [==============================] - 0s 2ms/step\n",
            "[[0 0]\n",
            " [0 1]\n",
            " [0 0]\n",
            " ...\n",
            " [0 0]\n",
            " [0 0]\n",
            " [0 0]]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "o0oyfLWoaEGw"
      },
      "source": [
        "### Making the Confusion Matrix"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Using confusion_matrix and accuracy_score\n",
        "\n",
        "from sklearn.metrics import confusion_matrix, accuracy_score\n",
        "cm = confusion_matrix(y_test, y_pred)\n",
        "print(cm)\n",
        "accuracy_score(y_test,y_pred)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "RwmRco9XOPCa",
        "outputId": "8a89b9e5-2ca3-4ac5-87da-76f8adfadedc"
      },
      "execution_count": 38,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[[1525   70]\n",
            " [ 207  198]]\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.8615"
            ]
          },
          "metadata": {},
          "execution_count": 38
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# CONCLUSION\n",
        "\n",
        "This ANN model helps the bank predict, with 86% accuracy, whether a customer is bound to leave or be retained by the bank."
      ],
      "metadata": {
        "id": "r5SEbdGARvFP"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Streamlit App"
      ],
      "metadata": {
        "id": "rg0w9lVwUCtr"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from keras.models import Sequential\n",
        "\n",
        "ann.save('customer_churn_model.h5')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Wlk0RkCyZyUB",
        "outputId": "7afdd077-7394-4ed4-82d4-41ed352d9b9e"
      },
      "execution_count": 47,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.10/dist-packages/keras/src/engine/training.py:3103: UserWarning: You are saving your model as an HDF5 file via `model.save()`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')`.\n",
            "  saving_api.save_model(\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Download the model file\n",
        "files.download('customer_churn_model.h5')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 17
        },
        "id": "vludSZK7aJ_n",
        "outputId": "272141e0-fad1-4bbd-b072-cf72ceee2096"
      },
      "execution_count": 48,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ],
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ],
            "application/javascript": [
              "download(\"download_e0d700da-df24-4484-a061-e0646cc1da63\", \"customer_churn_model.h5\", 68608)"
            ]
          },
          "metadata": {}
        }
      ]
    }
  ]
}