Classification of Credit Card Frauds Detection using machine learning techniques

Abstract

Credit card fraud refers to the illegal activities carried out by criminals. In this research paper, we delve into the topic by exploring four different approaches to analyze fraud, namely decision trees, logistic regression, support vector machines, and Random Forests. Our proposed technique encompasses four stages: inputting the dataset, balancing the data through sampling, training classifier models, and detecting fraud. To analyze the data, we utilized two methods: forward stepwise logistic regression analysis (LR) and decision tree analysis (DT), in addition to Random Forest and support vector machine. Based on the outcomes of our analysis, the decision tree algorithm produced the highest AUC and accuracy value, achieving a perfect score of 1. On the other hand, logistic regression yielded the lowest values of 0.33 and 0.2933 for AUC and accuracy, respectively. Moreover, the implementation of forest algorithms resulted in an impressive accuracy rate of 99.5%, which signifies a significant advancement in automating the detection of credit card fraud.