Precision in Fraud Detection involving Credit Card: A Comparative Study of Machine Learning Classifiers

Credit card transactions have become an integral part of modern commerce, but their increasing usage has led to a rise in fraudulent activities. To find fraud with credit cards, this study explores many methods of teaching machines, in particular K-Grand neighbours (KNN), the vector machine (SVM) and logistical regression (LR). An analysis of a set of very unbalanced data transactions carried out by European card owners is part of the process. To increase the effectiveness of the model, pre -processing procedures include the choice of functions and removal of a duplicate. The assessment is carried out using several performance parameters, including as accuracy, accuracy, review, F1-indicator and AUC-ROC. With a precision of 94.30%, it was discovered by the most effective model, which is logistical regression. Additional information from categorization reports, confusion matrices, and correlation matrices confirms the model’s usefulness. The difficulties of computing efficiency and real-time deployment are also covered.