DETERMINING THE IMPACT OF A COURSE ON STUDENTS’ GRADE POINT AVERAGE USING EDUCATIONAL DATAMINING TECHNIQUES

Abstract

Nowadays, educational sectors are becoming the first place that generate massive data. As educational data is increased, an appropriate data analysis will be mandatory in order to predict the status of education in the future. The study aim to apply Educational data mining to determine the impact of a course on students’ Grade point average. The study focused on Haramaya University College of computing and informatics college (CCI). Five code related courses (such as Fundamentals of database systems, fundamentals of programming, data structure and algorithms, internet programming, and object-oriented programming,) were selected for the study. Haramaya university registrar office was the data source for this study. The collected dataset covers eight years of student academic data from 2003- 2010 E.C and the study uses Hybrid data mining process model and WEKA3.9.2,Rapidminer 9.10.0 and orange 2.7,Microsoft excel 2013 were used for data mining and to preprocess. Separate dataset was prepared for each course. Finally, 6493 instance and three attributes were selected for the analysis. In addition to this, the value of grade and Grade point average (GPA) attributes were transformed based on the university assessment system which is categorized as excellent, very good, good, and satisfactory and failure. In this study two analysis were conducted clustering and association rule analysis and K mean, Apriori and FP-growth algorithms were used in order to get the results. By using minimum support 2% and minimum confidence 60% different rules were obtained. Then the discovered rule and similarity pattern evaluated using support and confidence to evaluate the discovered rules and SSE for cluster similarity. The cluster analysis result shown similarity patterns of student course result and gender in Data structure and algorithm and object oriented programming. As we seen 75% male, 25% female in DSA and 75% male and 29% female in OOP score satisfactory grade.so the resulting student clusters give more information to the college to perform different activities to solve educational problems that are related to the courses. Analysis of association rule mining shown that Fundamentals of data base system, Fundamentals of programming and Object oriented programming had a close impact on students’ semester Grade point average with 91%, 80%, and 79.9% of confidence respectively. This study used only gender, grade, and GPA as an attributes, therefore it would be better to use students’ lower grade background information like performance, rank, family background, school type (private or government) etc. to get accurate impact of GPA