Prediction of Cumulative Grade Point Average: A Case Study

Cumulative Grade Point Average (CGPA) prediction is an important area for understanding tertiary education performance trend of students and identifying the demographic attributes to devise effective educational strategies and infrastructure. This paper aims to analyze the accuracy of CGPA prediction of students resulted from predictive models, namely the ordinary least square model (OLS), the artificial neural network model (ANN) and the adaptive network based fuzzy inference model (ANFIS). We have used standardized examination (Secondary School Certificate and High School Certificate) results from secondary and high school boards and current CGPA in respective disciplines of 1187 students from Independent University, Bangladesh from the period of April 2013 to April 2015. Evaluation measures such as- Mean absolute error, root mean square error and coefficient of determination are used as to evaluate performances of above-mentioned models. Our findings suggest that the mentioned predictive models are unable to predict CGPA values of the students accurately with currently used parameters.