Manual Label and Machine Learning in Clustering and Predicting Student Performance: A Practice Based on Web-Interactive Teaching Systems

Abstract

This study presents the Academic Investment Model (AIM) as a novel approach to predicting student academic performance by incorporating learning styles as a predictive feature. Utilizing data from 138 Marketing students across China, the research employs a combination of machine learning clustering methods and manual feature engineering through a four-quadrant clustering technique. The AIM model delineates student investment into four quadrants based on their time and energy commitment to academic pursuits, distinguishing between result-oriented and process-oriented investments. The findings reveal that the four-quadrant method surpasses machine learning clustering in predictive accuracy, highlighting the robustness of manual feature engineering. The study's significance lies in its potential to guide educators in designing targeted interventions and personalized learning strategies, emphasizing the importance of process-oriented assessment in education. Future research is recommended to expand the sample size and explore the integration of deep learning models for validation.