Implementation of student SGPA Prediction System(SSPS) using optimal selection of classification algorithm

Abstract

The students have setup their goals before starting their engineering studies. To achieve their goals they need to succeed their engineering examinations with good marks and sit in the competition to get good job. The knowledge regarding success rate of students and factors affecting their performance is hidden in educational data set. Extraction of knowledge using data mining techniques helps students to know their weakness and work hard to improve it. In this study the Student SGPA Prediction System(SSPS) is developed which uses rules extracted from the best algorithm among J48, LMT, Random Tree and REP Tree algorithms to predict SGPA of students in first six semesters. These four classification algorithms are compared by building student performance prediction model based on student's social conditions and previous academic performance using WEKA. The records of 236 computer engineering students at Punjabi University are used to build these models. REP Tree algorithm with average accuracy (61.70%) and minimum average error rate(0.3608) is found to be better than the J48, Random Tree and LMT algorithms.