Machine Learning for Predicting Students’ Employability

Abstract

Graduates' employability becomes one of the performance indicators for higher educational institutions (HEIs) because the number of graduates produced every year from higher educational institutions continues to grow and as competition to secure good jobs increases, it is significant for HEIs to understand the employability of graduates upon graduation and highlight the reasons. To predict students' employability before graduation, machine learning models were employed. These include logistic regression; decision tree, random forest, and an unsupervised clustering (K-Means) algorithm. This research, therefore, aims to predict the full-time employability of undergraduate students based on academic and experience employability attributes – including cumulative grade point average (CGPA), student industrial work experience scheme (SIWES), co-curricular activities, gender, and union groupings before graduation. Primary datasets of 218 graduate students in the last four academic calendar years (2016 – 2021) from the Computer Science Department of Federal University Dutse were rated. The results demonstrate that Random Forest Classifier predict students employability the best with an accuracy of 98% and f1-score of 0.99 compare to logistic regression and decision tree. Furthermore, using more students’ data with more attributes including academics and extracurricular activities can improve the models performance and predict students’ employability.