DATA ANALYTICS MODEL FOR ADMISSION OF STUDENTS IN NIGERIAN TERTIARY INSTITUTIONS

ABSTRACT

This study is centered on the development of a data analytics model for admission of applicants into Federal University Lokoja for the purposes of making admission decision executed quickly, accurately and devoid of undue human interference or bias. We developed the model as a computerized system for the admission placement, which has been partly computer system and partly manual. We adopted the Cross Industry Standard for Data Mining (CRISP-DM) methodology because of the data driven nature of the study and compared seven (7) supervised learning classifier algorithms to determine that Support Vector Machine is also suitable for this kind of model using the kernel-tricks, as against other researchers claims, that Random Forest is most suitable for this kind of model. In our findings, the Gradient Boosting, Random Forest, Logistic Regression, Decision Tree, Naïve Bayes and Support Vector with the Linear Kernel-trick have 100% accuracy, 0% error rate, 100% precision, 100% Recall and 100% F1 score. We emphasized on Support Vector Machine Classifier, as it has great accuracy and precision, and also attempt to separate the target classes with the widest possible margin. The final output of the model clearly shows that engaging the model in decision-making in admission process will improve or yield good results by the admissions’ division of Federal University Lokoja.

Keywords: Data Analytics, Model, Admission, Decision Support, and Classification Algorithm.