Credit Card Fraud Detection using Hidden Naive Bayes and Bayesian Belief Network

This paper evaluates credit card fraud detection models using Hidden Naive Bayes (HNB) and Bayesian Belief Network (BBN) approaches. It aims to design accurate models for identifying fraudulent transactions in a dataset with various transaction features. HNB and BBN use probabilistic reasoning and statistical analysis to classify transactions as fraudulent or not based on these features. Performance metrics are employed to evaluate the models. Results show HNB achieving 86.87% accuracy, and BBN reaching 89.59%. Additionally, a comparison with other fraud detection approaches highlights HNB and BBN’s competitive performance. Overall, this paper showcases their potential in accurately detecting credit card fraud, offering valuable tools for this endeavor.