Fraud Detection on Smart Cards Using Machine Learning Algorithms

Authors

  • B. Rama Devi, K. Sri Harsha, Y. Himaja, B. Nagendra Babu

Abstract

One of the toughest problem in financial services is smart card fraud. Every year millions of dollars are going to be lost due to smart card fraud [4]. In recent times online transactions had become one of the most important part of our lives. Due to increase in number of transactions the fraudulent transactions [5] are also increasing rapidly. The main aim of this paper is to find out the finest and accurate model to detect the smart card fraud. Here some of the previously implemented machine learning algorithms [3] are chosen. Among those the top techniques that gives maximum accuracy levels are selected. In order to work on these algorithms the datasets that contains previous smart card transactions [4] are used. Some of the data pre-processing and data normalization techniques are applied on this raw data. To detect and reduce the fraud some of the machine learning algorithms like logistic regression, decision tree, support vector machine, k-nearest neighbour etc., are used. Among these decision tree provides more accuracy rate than the other algorithms and is stated as best for smart card [3] fraud detection.

Downloads

Published

2020-05-10

Issue

Section

Articles