A Model for Predicting Students’ Academic Performance using a Hybrid of K-means and Decision tree Algorithms

Abstract

Higher learning institutions nowadays operate in a more complex and competitive due to a high demand from prospective students and an emerging increase of universities both public and private. Management of Universities face challenges and concerns of predicting students’ academic performance in to put mechanisms in place prior enough for their improvement. This research aims at employing Decision tree and K-means data mining algorithms to model an approach to predict the performance of students in advance so as to devise mechanisms of alleviating student dropout rates and improve on performance. In Kenya for example, there has been witnessed an increase student enrolling in universities since the Government started free primary education. Therefore the Government expects an increased workforce of professionals from these institutions without compromising quality so as to achieve its millennium development and vision 2030. Backlog of students not finishing their studies in stipulated time due to poor performance is another issue that can be addressed from the results of this research since predicting student performance in advance will enable University management to devise ways of assisting weak students and even make more decisions on how to select students for particular courses. Previous studies have been done Educational Data Mining mostly focusing on factors affecting students’ performance and also used different algorithms in predicting students’ performance. In all these researches, accuracy of prediction is key and what researchers look forward to try and improve.