Comparative Performance Analysis of Machine Learning Models for Financial Fraud Detection

Abstract

Financial institutions increasingly rely on intelligent decision support systems to accurately identify high-risk credit applicants while minimising financial losses. However, credit risk prediction remains a challenging task due to the heterogeneous nature of customer information and the imbalance between good and bad credit classes. This study presents a comprehensive comparative analysis of supervised machine learning algorithms for credit risk prediction using the German Credit dataset. The proposed framework incorporates systematic data preprocessing, including missing value treatment, categorical feature encoding, feature scaling, and feature engineering, followed by model development under identical experimental conditions. Nine supervised classifiers, namely Logistic Regression (LR), Decision Tree (DT), Random Forest (RF), K-Nearest Neighbors (KNN), Naïve Bayes (NB), Gradient Boosting (GB), AdaBoost, Stochastic Gradient Descent (SGD), and Extra Trees (ET), were evaluated using multiple performance metrics, including Accuracy, Precision, Recall, F1-score, Receiver Operating Characteristic–Area Under the Curve (ROC-AUC), Precision–Recall curves, and confusion matrix analysis. Experimental results demonstrate that ensemble learning techniques consistently outperform conventional machine learning models on the imbalanced credit dataset. Among all evaluated classifiers, the Extra Trees model achieved the best overall performance with an Accuracy of 99.96%, Precision of 0.94, Recall of 0.74, F1-score of 0.83, ROC-AUC of 0.997, and AUPRC of 0.889, indicating superior discrimination capability and robust generalisation performance. Comparative analysis with representative studies further confirms the effectiveness of the proposed approach. The obtained results demonstrate that randomised ensemble learning provides an effective and computationally efficient solution for intelligent credit risk prediction and financial decision support systems.