Credit card fraud prediction and classification using deep neural network and ensemble learning

Abstract

The use of credit card has increased dramatically as a mode of payment in current times. As the number of credit card user is rising the frauds and identity thieves are also increasing. However, the prediction of if a transaction is faulty is a challenging task for banks. In this manuscript we are comparing predictive accuracy of the customer's default payment using Ensemble learning. We are going to create a model and use that model to predict which transaction is faulty. Moreover, we have applied four algorithms (Naïve Bayes Classifier Algorithm, Logistic Regression, Decision Trees and Deep Belief Network) to the dataset and then used ensemble learning to get the final result. We compare our study with the result of ensemble learning using three algorithms (Naïve Bayes Classifier Algorithm (Gaussian), Logistic Regression, Decision Trees) applied to the same dataset. The results of this paper will indicate the significance of deep belief network algorithm and that our proposed model which is ensemble learning combining the four algorithms perform better in predicting the default of credit card clients and portrays extreme precision.