Evaluation of Supervised Machine Learning Algorithms for Credit Card Fraud Detection: A Comparison

Abstract

Every day the modern world is moving towards digitalization and cashless transactions are becoming more common, credit cards are rapidly becoming more popular. Online and offline purchases using credit cards have become increasingly popular, which results in more fraudulent transactions every day. A large number of credit card fraud incidents occur every year and lead to huge financial losses. accordingly, it could be important to choose the best fraud detection method is essential so that it can detect fraud before criminal consumers a stolen card. To detect fraud, one method is to evaluate historical transaction data, as well as both normal and fraudulent transactions, to obtain usual and fraudulent behavior features by using machine learning techniques. we can use machine learning algorithms to solve this problem if we have access to enough data. In this study, our goal is to compare three algorithms for detecting credit card fraud (Decision Tree, Regression Logistic and Random Forest). we want to use a model that is new and based on a hybrid approach for detecting credit card fraud. According to this study, the proposed model is more capable of identifying fraudulent transactions than previous studies.