Feature engineering strategies based on a One-point Crossover for fraud detection on Big Data Analytics

Abstract A wide range of new opportunities for fraudulent online activities has arisen with the growing popularity of online shopping and big data issue. E-payment fraud schemes are collecting billions of dollars from customers, distributors and service providers every year. A lot of machine learning methods for fraud detection problems have been proposed which can be categorized into supervised, unsupervised and semi-supervised methods. In this paper, we proposed biologically inspired technic in the feature engineering phase for handling imbalanced data to increase the total data of a small number of classes by oversampling. One-point crossover used to generate the new data of minority classes. The best algorithm performance obtained to predict the fraud transaction from various machine learning models is Classification and Regression Tree with the corresponding accuracy, precision, recall, and F-1 Score are 96%.