Credit Card Transactions Fraud Detection for Multiple Consumer Behaviors

In today's digital era, credit card fraud is a preva-lent issue that costs financial institutions and individuals billions of dollars. To prevent such fraud, fraud detection systems are implemented that use machine learning algorithms to analyze patterns and detect transaction anomalies. However, these systems heavily rely on historical data, and if the data is limited or biased, the system's accuracy decreases significantly. This study addresses this issue by investigating a real credit card transaction dataset and determining different consumer behaviors. Synthetic datasets are generated based on consumer behaviors to enhance the accuracy of detecting fraudulent activities. According to the findings, the Logistic Regression (LR) model exhibited superior performance in both experiments. It achieved an impressive accuracy of 96.4% with remarkable time efficiency in the first experiment, and 94.5% accuracy in the second experiment, while still maintaining excellent time efficiency. This research goes through the procedures involved in analyzing a real dataset, understanding consumer behaviors, and generating synthetic datasets.