Predicting Students’ Graduating Cumulative Grade Point Average Using Difference Level, Classification and Regression Tree and Linear Regression Algorithm


  • journal manager


CART, Difference level (DL), Linear regression, Prediction, Student graduating CGPA


Azeez, T.O., Awe, A. C. and Omosebi, P. A.
Department of Computer Science,
College of Natural Sciences,
Joseph Ayo Babalola University,
Ikeji-Arakeji, Osun State, Nigeria.
Department of Computer Science,
Faculty of Science,
University of Lagos,
Akoka, Lagos, Nigeria.

Predictive modeling using data mining methods for early identification of students’ performance can be very beneficial to
forecasting students’ graduating class of degree. It is an innovative methodology that can be utilized by universities. The goal of
the study is to have an early detection of graduating students’ cumulative grade point average (CGPA) before they eventually
graduate. Classification and regression tree (CART) and linear regression were the algorithms used to carry out the prediction
model. Also, a novel algorithm: Difference Level (DL) was designed and incorporated in the system. The system works by
taking the differences of each level grade point average and adding the resultant values together; then subtracted from their final
year first semester result to give a predicted graduating cumulative grade point average. Data analysis was performed on datasets of a specific graduated class. The dataset was obtained from Joseph Ayo Babalola University (JABU) Exams and Records unit. This study found that students with a risk of graduating with a low CGPA can actually be predicted at the end of the final year
first semester. The aim of this study is to help students in improving their ability in getting a better graduating CGPA.