Abstract:
This research presents a method for selecting the factors which affect the achievement of students grade called the Dynamic Feature Selection Optimization of Subspace Classification Algorithm (DFS-SCA). We set up the learning management system (LMS) for the online course named the application of computer in daily life, as well for other blended courses, to collect the data of all 1,605 students throughout 6 semesters (2/2010 to 1/2013). The evaluation results revealed that with Subspace Clustering Algorithms (SCA) the accuracy of up to 93.64% could be achieved. The factor which was most subject to this significant efficacy wasFormative test
lesson 5,final exam scores, pre-test exam scores, post-test exam scores, and e-learning scores. Considering the efficiency of each subspace using the Correctly Clustered Instances (CCI) values, it was found that scores of level A had highest value of 100%. To cluster the students, 5 factors were used consisting of Pre-test, Post-test, average scores of all tests for the entire semester, number of passed tests during the semester, and scores of the final examination.