AI-Powered Career Guidance System: A Machine Learning Approach Using Holland's RIASEC Model for Personalized Occupation Recommendations

Abstract

Career guidance continues to be a challenge to students and professionals searching for optimal vocational placement. Current methods of counselling use manual analysis of psychometric assessment answers by trained counsellors, which is not scalable and subject to subjective bias. This paper proposes an end-to-end machine learning-based career guidance system that incorporates Holland's RIASEC vocational interest theory into a web-based system capable of recommending 923 ONET occupations from a 48-item psychometric assessment and providing plain-explained reasoning. Decision Tree, Support Vector Machine (SVM) and XGBoost regressors were trained on 8,855 real responses from Open Psychometrics and compared with a traditional Pearson-correlation baseline profile matching method. Evaluation on three complementary metrics—Hit Rate, Euclidean Distance, and Profile Correlation—shows that ML-augmented methods significantly outperform the baseline, with SVM achieving an 81.93% Hit Rate compared with 34.68% by the Profile Method. A recommendation API, an interactive web interface with radar-chart visualization, and explainability of occupation recommendations via dimension overlap are proposed. This work bridges vocational psychology with practical AI engineering, providing a scalable and interpretable solution for career counseling suitable for deployment in educational institutions.