Discovering vital patterns from UST students data by applying data mining techniques

Abstract

This paper presents an applied study in data mining and knowledge discovery. It aims at discovering patterns within historical students' academic and financial data at UST (University of Science and Technology) from the year 1993 to 2005 in order to contribute improving academic performance at UST. Results show that these rules concentrate on three main issues, students' academic achievements (successes and failures), students' drop out, and students' financial behavior. Clustering (by K-means algorithm), association rules (by Apriori algorithm) and decision trees by (J48 and Id3 algorithms) techniques have been used to build the data model. Results have been discussed and analyzed comprehensively and then well evaluated by experts in terms of some criteria such as validity, reality, utility, and originality. In addition, practical evaluation using SQL queries have been applied to test the accuracy of produced model (rules). ©2010 IEEE.