Performance Analysis of Machine Learning Models and Deep Learning Model for Credit Card Fraud Detection

Credit card fraud remains a prevalent issue, causing significant financial losses that impact both institutions and their customers. Conventional approaches to detecting fraudulent activities struggle to identify increasingly sophisticated and adaptive fraudulent activities, highlighting the demand for more advanced solutions. This study examines the performance of several supervised machine learning algorithms for detecting fraudulent credit card transactions. Nine models are implemented and compared, including Logistic Regression, K-Nearest Neighbors, Decision Trees, Random Forest, as well as ensemble methods such as AdaBoost and Gradient Boosting. Key performance metrics, including precision, recall, F1 score, accuracy, and overfitting tendency, are analyzed using a labeled dataset of 10,48,575 transactions to assess each model's effectiveness. Results indicate that ensemble models, particularly Random Forest and Gradient Boosting, achieve superior accuracy and robust generalization in fraud detection. By employing these advanced machine learning methods, financial institutions can enhance their ability to identify fraud, reduce financial losses, and increase security for legitimate customers.