Credit Card Fraud Detection Using Supervised Learning Algorithms

Different machine learning techniques demonstrate exceptional performance in numerous banking challenges, with one of them addressing the ever-present issue of credit card fraud detection. This paper examines the performances of Logistic Regression, Random Forest, and CatBoost in the context of credit card fraud. The capability to handle imbalanced classes is evaluated for each algorithm, as well the impact of hyperparameter tuning on model performance. The findings reveal that algorithms based on decision trees effectively manage imbalances without need for additional data preprocessing. CatBoost outperforms other algorithms in all standard metrics, making it the most desirable choice for addressing this specific problem.