PREDICTING GRADUATE EMPLOYABILITY USING ENSEMBLE LEARNING ON RESUME DATA, INTERNSHIPS, AND SOFT SKILL SCORES

Abstract

Graduate hiring has become an important way to judge how well higher education systems work, so it's important to use data-driven methods to correctly predict job results. This study suggests using ensemble learning to guess how likely it is that a graduate will be able to get a job after graduation. It does this by combining different types of data, such as resume material, internship experience, and soft skill assessment scores. Natural language processing (NLP) methods are used to pull out meaning information from resume data, like school successes, professional skills, certifications, and leisure activities. Some aspects of internships, like length, importance to the job, and company rank, are measured and used as number forecasts. Psychometric scores and friend evaluations are used to standardize and add to the feature area the level of soft skill proficiency. The forecast model uses a group of ensemble learning methods, such as Random Forests, Gradient Boosting Machines (GBM), and Extreme Gradient Boosting (XGBoost). It also uses a meta-learner that is applied through stacking to improve its ability to generalize. Feature importance analysis shows how important cognitive, behavioral, and practical traits are in terms of finding key employment markers. Cross-validation and hyperparameter tuning are used to make sure that the model is stable, and the ensemble model performs better than individual classifiers in terms of accuracy, precision, recall, and F1-score. This work not only creates a scalable method for predicting how employable graduates will be, but it also gives schools useful information that they can use to improve their courses and gives students useful information that they can use to customize how they prepare for careers.