Objective: To build a model to predict the fraudulent transaction, so that we can stop payment of such transaction and save customer money.
Challenges: Dataset was highly imbalanced. Except two variables, all the variables were the result of a PCA transformation and contain only numerical input.