Utilizing Data Science and Analytics in Predicting Campus Placement

Abstract

Educational institutions are expected to make students marketable in their respective fields. Job placement exam is a tool to assess a student's readiness to face the industry's challenges. Previous studies have utilized machine learning algorithms to predict students' job placement. However, most of the past research was based on academic and non-academic performance metrics, not on a custom-made job placement exam. The training and test data used in the research were from computer science engineering students who took a job placement exam. The study examined the scores of job placement exam in the different subject areas. In this study, five machine learning methods were utilized to develop the predictive models. Of the five models explored, the random forest model got the highest accuracy, 90.85%, and an F measure of 91.59%. Feature selection using a forward algorithm was then employed to get the most influential predictors. The results showed that coding was deemed to be most important, followed by the aptitude score.