An efficient scheme for credit card fraud detection system using decision tree for streaming transactional data in comparison with hmm based automated feature engineering

To perform innovative fraudulent transaction detection in credit card using decision tree algorithm and its performance is tested by comparing with the hidden markov model (HMM) in machine learning. Materials and Methods: Decision tree classifier and hidden markov model algorithm with sample size N=10 in each group is iterated 10 times for credit card fraud detection using innovative extract transaction behavioural pattern by comparing with the metrics like accuracy, specificity and f-score. Results: The detection accuracies of decision tree algorithms and hidden markov model are 97.8% and 94.9% respectively. Conclusion:Results demonstrate that the proposed decision tree algorithm gives significantly better performance than the hidden markov model algorithms in innovative fraudulent transaction detection in credit card.