Hybrid Machine Learning Algorithm for Credit Card Fraud Detection

Abstract

The prediction analysis (PA) refers to an approach using which the upcoming prospect can be predicted from the previous occurred events. This technique has two phases namely feature extraction and classification process. The detection of fraud in credit card using PA becomes challenging because of the complexity in datasets. There are numerous classification methods that are implemented in state-of-art schemes in order to detect the frauds in credit card. The prediction analysis is a DM (data mining) method which is useful for future forecasting on the basis of current information. This research is carried out to perform the CCFD (credit card fraud detection) on the basis of recent information. The data of credit card is available in an enormous volume form. Consequently, it becomes difficult to establish association among diverse features that have impact on the predictive accuracy. The KNN (K nearest Neighbor) is deployed to extract the attributes and the PA (prediction analysis) is performed by the means of NB (naïve bayes) algorithm.