Comparison of Supervised Machine Learning Approaches in Predicting Employability of Students

Abstract

This study investigates the efficacy of various supervised machine learning approaches in predicting the employability of graduates using data from mock job interviews conducted in multiple universities and agencies across the Philippines. Employing a dataset comprising 2,982 observations, the research aims to determine which machine learning algorithm—NaiveBayes, Logistic Regression, k-nearest neighbors (IBK), or Random Tree—yields the most accurate predictions. Additionally, the study assesses the impact of eight non-cognitive factors on employability: general appearance, manner of speaking, physical condition, mental alertness, self-confidence, ability to present ideas, communication skills, and student performance rating. Using WEKA software, the analysis reveals that the Random Tree algorithm provides the highest prediction accuracy, followed by k-nearest neighbors. The findings underscore the significance of non-cognitive skills in graduate employability, with general appearance, manner of speaking, and physical condition emerging as the top predictors. This research highlights the potential of machine learning in enhancing employability outcomes by identifying critical attributes that influence job market success.