Development of a Mobile-Based Machine Learning Prototype for University Admission Suitability Prediction: A Case Study in an Information Technology Program

Abstract

This study presents an end-to-end machine learning framework for predicting admission suitability to an Information Technology program, designed with an emphasis on interpretability and real-world deployment. The proposed approach integrates imbalanced-data handling using SMOTE, a Random Forest classification model, and a mobile-based decision-support system implemented via a Spring Boot API and a cross-platform Flutter application. Using real academic data from high school students, the system predicts admission suitability based on subject-level performance and background attributes. Experimental results demonstrate that the proposed model achieves balanced classification performance, particularly in improving the detection of underrepresented cases, while maintaining stable performance across validation procedures. Unlike prior work that primarily focuses on predictive modeling in isolation, this study contributes a unified framework that combines interpretable machine learning with a deployable system accessible to students and academic advisors. The integration of model explainability and mobile deployment enables transparent and practical decision support, allowing users to understand both prediction outcomes and the key factors influencing them. These findings highlight the potential of combining machine learning with system-oriented design to support informed decision-making in educational contexts.