Poor academic performance has discouraged and made so many students drop out of university due to National University Commission (NUC) regulation of 1.5 probational level which consequently affects list of enrolment in higher institutions yearly. Numerous researchers have worked on this over the years in order to proffer solution to this but suffer a limitation of fewer accuracy level. Hence, this research was developed to proffer solution to the limitation by developing a result prediction system for undergraduates using data mining algorithms. In this research work, 448 instances of datasets were collected which consist of students Secondary School Certificate Examination (SSCE) results, student secondary school result in five major subjects, Joint Admission Matriculation Board (JAMB) score, and Cumulative Grade Point Average (CGPA) at their first year in the university. The machine learning algorithms employed in the research were K-Nearest Neighbor, Random Forest, J-48, and Naïve Bayes from WEKA (Waikato Environment for Knowledge Analysis). From experimental results obtained, the most valid data that affects the CGPA of an undergraduate student are the combination of JAMB result with other features such as Age and other SSCE result. The accuracy result of the developed system, using the selected algorithms are as follows: K-Nearest Neighbour (81.40%), RandomForest (85.26%), J-48 (79.65%), Naïve Bayes(76.14%). Therefore, based on the result of analysis, it was discovered that Random Forest outperformed all other machine learning algorithms. The results from this
research can be employed by career advisors and Educationalists to advice students as appropriate. This research has been able to use real-world data to evaluate the success of students in tertiary institutions with the use of machine learning approach.