Design of a Data Warehouse-Based Early Warning System for Student Drop Out Prevention

Abstract

Higher education institutions require integrated, analytics-based data management to support strategic decision-making and student drop out prevention. This study aims to design a Data Warehouse (DW) model as the foundation for an Early Warning System (EWS) to detect student drop out risks at Universitas Paramadina. The DW is designed using the Kimball lifecycle approach with a star schema implementation, integrating data from multiple business processes such as academics, finance, and LMS activities. The EWS is developed using a supervised learning classification approach, utilizing Logistic Regression as the baseline model and proposing Random Forest for advanced modeling. The results demonstrate that an integrated DW effectively supports machine learning-based predictive analytics and serves as a strategic framework for proactive student drop out prevention.