Abstract

Credit card payment is gradually becoming a most preferred mode of payments globally. Like many new innovations that invariably turns out to be a success story economically, it often passes through a difficult development stage and acceptance by users due to social and cultural reasons. Implementing Credit/Debit card payments have security challenges, a bane for low acceptance by users from social and cultural standpoints. Big data analytics is one way by which the world is solving and coping with this challenge. The rate at which fraudsters use credit/debit card to commit crimes is on the rise and this need to be curbed to the barest minimum to leverage on the benefits that technology has brought. Though, there are different models for curbing the growing trend of credit card fraud informs of identifying and isolating scenarios, but these are incapable of dealing with the trends. This paper proposes an Ensemble model (that is, Random Forest Classifier, Gradient Boosting Classifier, and CatBoost Classifier) to identify patterns and fraudulent activities in transactions made using credit cards. The dataset of credit card transaction was used to monitor the transaction behaviour of credit card owners. The ensemble model was trained on the outputs of the three individual machine learning classifier algorithms using the stacking classifier. The results showed that, the model achieved prediction accuracy of 96%, a precision score of 98% for the fraudulent transactions, and 96% for non-fraudulent transactions.