ANN model build to predict a bank's customer churn.
https://parisrohan.medium.com/bank-customer-churn-prediction-using-ann-6499bf805b6
- The dataset is downloaded from Kaggle using this link: https://www.kaggle.com/datasets/mathchi/churn-for-bank-customers
- The dataset contains 1000 records and 13 features with no missing values.
- This is an imbalanced dataset as the number of non-churned customers is greater than the number of churned customers.
- The number of male customers is slightly greater than the number of female customers for the given bank.
- Around 50% of the customers are from the France region and the number of customers from the Spain and Germany region is almost equal.
The data preprocessing includes applying one-hot encoding on the Geography and Gender features.
- The data is scaled down using StandardScaler after splitting it into train and test data.
- Defining ANN
- Input Layer — As the training data has 11 features, the input layer will have 11 neurons.
- Hidden Layer — This depends on the trial. I have chosen 3 hidden layers.
- Output Layer — As this is a binary classification problem a single neuron will work in the output layer.
- ReLU activation function is applied on the hidden layers and Sigmoid function is applied on th output layer.
- Optimizer used is Adam optimizer.
- The binary cross-entropy loss function is used as this is a binary classification problems.
- Early stopping is used to stop the Neural Network if there is no significant improvement in the model’s accuracy.
- The model scores an accuracy of 85%
