Contrastive Study of Machine Learning Techniques for Credit Card Fraud Detection
Dhwanir Shah, Lokesh Kumar Sharma · Indian Journal of Science and Technology · 2025
Objective: To assess the efficacy of five machine learning algorithms—Logistic Regression, Support Vector Machine (SVM), Decision Tree, Random Forest, and XGBoost—in detecting credit card fraud, utilizing a simulated Kaggle dataset created through Sparkov for credit card transactions. Methods: The dataset was partitioned into three training-test ratios: 60%:40%, 70%:30%, and 80%:20%. To address class imbalance, the Synthetic Minority Over-sampling Technique (SMOTE) was employed. The performance of the models was measured using Accuracy, Precision, Recall, F1-score, and ROC-AUC, with validation conducted through 10-fold cross-validation. Findings: XGBoost consistently surpassed the other models, achieving precision and recall rates of up to 99.05% and 99.81%, respectively. The Decision Tree and Random Forest models produced precision scores of 96.86% and 94.65%, with recall values of 99.08% and 99.49%. Logistic Regression and SVM exhibited comparatively lower performance across all training-test splits. Novelty: This study introduces a unique methodology by evaluating machine learning algorithms across different training-test splits and validating the outcomes through 10-fold cross-validation to ensure the robustness of the models. Keywords: Credit card Fraud, Logistic Regression, Support Vector Machine, Decision Tree, Random Forest, XGBoost, K-Fold cross validation, OneHotEncoding