A Preliminary Study on Prediction Models for English Web-based Remedial Education: Application of Data Mining Theory

Remedial course is a product of a well-developed higher education system. As universities/colleges become more available to high school graduates, competitiveness in the entrance exam is lessened and demand for students’ basic scholastic abilities diminishes in admission. This social phenomenon causes two problems for the implementation of higher education: (1) Students can’t comprehend lectures due to their lack of basic academic knowledge; (2) The unbalanced distribution of students’ scholastic achievements is believed to be the most significant factor weakening teaching effectiveness. Many universities/colleges, especially those in developed countries and areas, have to integrate remedial courses into their first-year education in order to help the students compensate for what they should have mastered in the previous learning. This study reports an English remedial course carried out for all freshmen in Nishinippon Institute of Technology, Japan, and focuses on the regression analysis for determinants of students’ score changes. Two prediction models are postulated, so that the issue is being examined from different perspectives. In the first model, students’ total engagement in learning, intercultural communication competences and computer operation skills are proved to be responsible for their score changes. In the second model, students’ positive attitudes towards remedial course and webbased learning turn out to be a crucial factor to promote their performances. Key-Words: Remedial courses, Multiple regression analysis, Prediction model, English education, Web-based learning