An improved recommendation model using linear regression and clustering for a private university in Thailand
K. Kongsakun · 2013
In order to enhance the number of completions, educational institutes establish and implement strategies to improve students' satisfaction and academic development. Technological supports are a strategy that many universities provide to service and assist staff and students. In this study, a prediction model called Electronic Grade (e-Grade) is used to model the likelihood of a student's Grade and achievement for particular subjects. This model aims to assist lecturers to supervise students, and to pay extra attention for students who are likely to get marginal results that could lead to withdrawal prior to completion of that subject. The e-Grade model comprises two sub-models that are Likelihood of Grade Before midterm examination, and Likelihood of Grade After midterm examination. In the experiment, two datasets are used. The results will provide supports to counsel the students on their possible performance and classroom achievement. The usefulness of the proposed e-Grade model for the monitoring of students' progress is verified against benchmark data. The results are interpretation of the performance of the new model of research based on linear regression and clustering techniques. The experiment results found that the proposed model enhances the accuracy of linear regression techniques in comparison to the benchmark model.