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