Predicting student admission decisions by association rule mining with pattern growth approach

Abstract

Association Rule Mining is the most powerful technique in data mining. Generation of rule involves two phases where first phase finds the frequent itemsets and second phase generates the rule. Many algorithms are specified to find frequent item set from the sequential patterns. There are mainly two approaches for finding frequent item sets. First approach is with candidate sequence generation i.e. Apriori approach and second with the pattern growth method. Experiments shows that if the length of sequence is small, pattern growth method performs better than that of Apriori approach. In this paper we have analyzed the pattern growth method for the dataset of engineering student. With finding associations among the attributes we can find the chances of taking admission and predict the admission decisions of students. To find strong and valid association rules, different measures like min Interest, lift, leverage and conviction are considered. Prediction is achieved with the use of consequence constraint during generation of association rules.