An Efficient Approach for Credit Card Fraud Identification with the Oversampling Method

The financial business is rapidly expanding, and as a result, banking online transactions are increasing as the government pushes digital transactions. The bulk of financial transactions has been made with debit or credit cards. As a result, the amount of fraud linked with it is increasing. However, because existing fraud detection machine learning algorithms are trained and subsequently assessed on severely uneven datasets, their performance in real-world situations suffers. We presented a method in this study that may operate better after turning these unbalanced datasets into balanced datasets using the oversampling approach, ensuring that the system is not biased when the algorithm is used. The results show that oversampling method with the Random Forest algorithm performs better than the Logistic Regression algorithm with the oversampling method with 80% of accuracy.