Predicting Faculty Career Progression in Higher Education: A Random Forest and Permutation Importance Approach

This study investigates the determinants of faculty career progression in higher education through a machine-learning framework combining Random Forest (RF) classifiers and permutation feature importance analysis. Leveraging institutional data on faculty demographics, educational backgrounds, and career histories, we preprocess unstructured data via rigorous techniques such as one-hot encoding, and temporal feature engineering. The RF model, optimized with balanced class weights and bootstrap aggregation, demonstrates robust predictive performance. Permutation importance analysis identifies highest education (importance score: 0.2105) as the dominant predictor, underscoring the institutional prioritization of advanced degrees in promotion systems. While age (0.1210) and gender (0.0974) exhibit significant influence, the minimal impact of work experience (0.0684) suggests latent biases favoring chronological age over explicit tenure. Notably, ethnicity shows negligible relevance (0.0026), aligning with equity policies. Methodologically, this work validates permutation importance as a model-agnostic tool for auditing fairness in educational HR systems. The findings urge universities to re-examine promotion criteria by integrating competency-based metrics and addressing implicit age-gender disparities. This study contributes to educational technology by bridging predictive analytics with policy design for equitable academic career development.