A Comparative Analysis of Machine Learning Models for Credit Card Fraud Detection

The rapid growth of online transactions has greatly increased the autonomy of credit cards, creating serious challenges for financial institutions and consumers.Credit card fraud detection involves identifying unauthorized transactions conducted using stolen or fake credit card information. This study aims to explore the most influential features contributing to credit card fraud and evaluate the effectiveness of various machine learning models in predicting fraudulent transactions. Utilizing the "creditcard.csv" dataset, which contains real-world credit card transactions, we conducted feature selection and model comparison to enhance detection accuracy. The results demonstrate that Neural Networks and Support Vector Machines (SVM) are the most effective models, achieving high Matthews Correlation Coefficient (MCC) scores due to their ability to handle high-dimensional data and complex nonlinear relationships. In contrast, simpler models like Naive Bayes and Random Tree exhibited lower performance but can be improved through advanced techniques such as feature selection and data balancing. These findings highlight the importance of robust feature engineering and careful model selection in developing accurate and reliable fraud detection systems. Financial institutions can drastically improve their fraud detection capabilities by putting these innovative approaches into practice.