Hybrid Stacking Algorithm to Detect Fraudulent Transactions in Credit Card

With the advancement of technology, cyber crimes are increasing more. Many of the cyber crimes are based on the credit cards because with the invention of WIFI credit cards, the frauds are becoming easy with no OTP system. A model is required which can identify the unauthorized or outlier transactions using machine learning approaches. Researchers has implemented traditional and ensemble algorithms for identifying the unauthorized transactions. Few have implemented clustering techniques to recognize the outliers in the transactions but both of them are failed because of the more misclassifications and wrong assumptions of the parametric values in the clustering algorithms. So, in this the proposed model implements two level architecture (Stacking) in which lower level known as “base classifiers” are implemented using the boosting algorithms and at the second level known as “meta classifiers” are designed using the regression models to pick the one with majority voting.