Design the Modified Multi Practical Swarm Optimization to Enhance Fraud Detection


Financial fraud remains an ever-increasing problem in the financial industry with numerousconsequences. The detection of fraudulent online transactions via credit cards has always beendone using data mining (DM) techniques. However, fraud detection on credit card transactions(CCTs), which on its own, is a DM problem, has become a serious challenge because of twomajor reasons, (i) the frequent changes in the pattern of normal and fraudulent online activities,and (ii) the skewed nature of credit card fraud datasets. The detection of fraudulent CCTs mainlydepends on the data sampling approach. This paper proposes a combined SVM- MPSOMMPSOtechnique for credit card fraud detection. The dataset of CCTs which consists of284,807 transactions performed by European cardholders in 2013 was used in this study. Theproposed technique was applied to both the raw dataset and the pre-processed dataset. Theperformance of these techniques is evaluated based on accuracy, and the fastest time it takes todetect fraud. This paper, proposed a technique that uses SVM, MPSO and MMPSO to form anensemble for the detection of credit card fraud