/Bank_binary_classification

This is a binary classification problem: will a client leave the bank or stay

Primary LanguageJupyter NotebookMIT LicenseMIT

Bank_binary_classification

This is a binary classification problem solved with ANN. We have data provided by a bank that wants to predict if clients will likely to leave the bank (stop using services) or will continue using them.

We have used ANN for this problem with 2 dense (fully connected) layers of 6 units (neurons) each for imput and one activation layer with 1 unit for output.

We used cross-validation to improve our accuracy and early stopping in order to avoid model overfitting.

Dataset has been taken from https://www.kaggle.com/yogirj/churn-modelling Thank you to Kaggle and anyone behind this dataset.

In the end we were changing hyperparameters empirically in order to improve accuracy of the model