Application of Machine Learning in Prioritizing the Components of Islamic Financial Management

Abstract

Prioritizing the influential components of Islamic financial management is one of the fundamental challenges facing researchers and policymakers in this field. Traditional methods are often based on expert judgment and qualitative approaches and are subject to limitations such as bias and lack of reproducibility. The present study aimed to provide a quantitative, data-driven approach for prioritizing the components of Islamic financial management using the XGBoost algorithm and the SHAP method. Data were collected from 382 questionnaires and analyzed using k-fold cross-validation. The findings showed that the XGBoost model achieved an accuracy of 89.2% and an AUC-ROC of 92.4%, indicating excellent predictive performance. SHAP-based prioritization showed that “distributive justice” (0.452), “prohibition of riba” (0.381), and “information transparency” (0.347) had the highest priorities, whereas “accountability to investors” (0.189) and “moderation in consumption” (0.207) had the lowest priorities. The findings demonstrate that a machine-learning approach can be used as a complementary tool and, in some cases, as an alternative to traditional prioritization methods in Islamic finance and can contribute to greater accuracy and objectivity in managerial decision-making.