Hybrid rule based machine learning approach on forecasting students intake.

This paper proposes a hybrid rule-based machine learning approach to produce a student profiling model for University Selangor (UNISEL) a state-owned private higher education institution in Malaysia. This project attempts to address UNISEL's interest by proposing a student profiling model that can assist in predicting student intake using existing data from a survey conducted by UNISEL among its current students. Maintaining a healthy student intake is vital to ensure the continuity of any private educational institution. Hence, this research aims to predict student intake towards academic entrance in UNISEL based current data profile using hybrid rule-based methods subject to past student intake data to improve better decision-making process. The two algorithm that had been selected using the hybrid rule-based machine learning approach combines rule-based (i.e., expert system) with multi-level machine learning (i.e., K-Mode's clustering and decision tree classification) to produce the final student profiler model. This paper first outlines the assumptions made before dividing the proposed approach into five main phases: data cleaning, feature extraction and selection, model formulation and model testing. Results shows that the project managed to propose a creative workaround that suited the use-case presented to produce the student profiler model in predicting student intake in UNISEL. [ABSTRACT FROM AUTHOR]