A model for forecasting cumulative grade point average score (performance)

Abstract

This study identified the challenges of dwindling academic performance by students, the dilemma of course advising and the absence of early warning signals regarding student‟s academic performance. The study observed that the course advising dilemma, and the early warning signal absence derives from overwhelming student‟s enrollment into some programs, especially in public universities, and the absence of an easy to use system for projecting performance. To address these challenges, the study conceptualized the development of an easy to use Cumulative Grade Point Average (CGPA) forecasting system from existing student‟s records using data mining techniques. The Waikato Environment for Knowledge Analysis (WEKA) data mining tool with some of its inbuilt algorithms (J48 decision tree, Naïve Bayes and Neural Network) were used on the study data set. The mean absolute error and the relative absolute error obtained for the three algorithms were J48 decision tree: 0.049 and 35.037, Naïve Bayes: 0.214 and 89.592, Neural Network: 0.092 and 72.171 respectively. As a result of its least error rate, the J48 decision tree algorithm was adopted and the pseudocode generated by it was coded using the Visual Basic dot Net (VB.Net) Integrated Development Environment (IDE). The successfully developed system was then tested on various existing results set, and the outcome showed that the desired features and functionalities were achieved.

Keyword: Data Mining, Students Performance, Course Advising, Results Forecast, Cumulative Grade Point Average