Real-time Fraud Detection in e-Market Using Machine Learning Algorithms

An electronic market (e-market) is an online platform where people buy or sell products. Problems like fraud detection and illegal activity have risen together with the rising growth of the e-market. The efficacy of the fraud prevention methods of purchases has a significant bearing on the depletion of internet customers. Therefore in this paper, a support vector machine-based fraud detection framework (SVM-FDF) has been proposed for detecting real-time fraud in the e-market. FD framework is implemented to spread prominence from a limited marketing scheme for beginning consumers is invariably used to update their credibility when an offering is applied to the e-market. The comportment features of all existing regular cases and fraud specimens are derived via the clustering algorithm to form the general conduct of the present community of the e-market. Each conduct's findings demonstrate that the SVM model is employed to evaluate whether all the present transaction is corrupted or fraud. The simulation results show that the suggested SVM-FDF model enhances the precision rate of 98.8%, recall rate of 97.7%, the f1-score ratio of 96.7%, accuracy ratio of 96.8%, and decreases the error rate of 20.9% compared to other existing approaches.