Attribute Selection on Student Academic and Social Attributes Based on Randomized And Synthetic Dataset

Subset selection is important when the underline dataset contains insignificant attributes as they don’t contribute much to the final results especially, in the context of student performance prediction studies. Hence the exploration of procedures for such goal becomes relevant. Though this is the case, in general it happens to be NP-hard problem. In this paper we apply Best First Search, Greedy Search, and Ranker method (Information Gain Ratio) to select the attributes using weka tool with learning models based on decision rules, decision trees, neural networks, bayes NET and meta classifiers. The performance comparisons are made with ranked and non-ranked search methods over the synthetic and randomized datasets derived from original students’ performance dataset.