Over the years, data mining has been successfully adopted to transform the business world by implementing different models for evaluating business performance. Analyzing and evaluating large volume of dataset by these data mining models keeps its application growing wide. Several data generated from student academic results and bio data calls for a need to create knowledge out of the data set. Students' academic performance evaluation is a necessity and incredibly challenging, and thus, is intended for identifying and extracting new and potentially valuable and actionable knowledge from the data. It is a complex research undertaking to identify and indicate the issues harming students’ academic performance. Performance prediction models can be built by applying machine learning tools to enrolment data. This paper presents five Machine Learning models-K-Nearest Neighbour Classifier, Random Forest, Gaussian Naïve Baye’s Classifier, Decision Tree, and Support Vector Classifier-for predicting students’ continuous performance and graduating cumulative Grade Point Average. Each model is applied to data set on the enrolment data and examination results for three different academic sessions ranging from the first year to the graduating year. The comparative analysis of the performance of each of these models is carried out based on accuracy of prediction.