An Efficient Study of Fraud Detection System Using Ml Techniques

Abstract

The growing world has the transactions of finance mostly done by the transfer of amount through the cashless payments over the Internet. This growth of transactions led to the large amount of data which resulted in the creation of big data. The day-by-day transactions increase continuously which explored as big data with high speed, beyond the limit of transactions and variety. The fraudsters can also use anything to affect the systematic working of current fraud detection system (FDS). So, there is a challenge to improve the present FDS with maximum possible accuracy to fulfill the need of FDS. When the payment is made by using the credit cards, there is chance of misusing the credit cards by the fraudsters. Now, it is essential to find the system that detects the fraudulent transactions as a real-world challenge for FDS and report them to the corresponding people/organization to reduce the fraudulent rate to a minimal one. This paper gives an efficient study of FDS for credit cards by using the machine learning (ML) techniques such as support vector machine, naïve Bayes, K-nearest neighbor, random forest, decision tree, OneR, AdaBoost. These machine learning techniques evaluate a dataset and produce the performance metrics to find the accuracy of each one. This study finally reported that the random forest classifier outperforms among all the other techniques.