Data Driven Student Institution Matching Evaluation Based on Random Forest and Causal Inference for Vocational Enrollment Systems
Year: 2025
Type: conference-paper
Study design: Observational Study
Abstract: In the era of data-driven decision-making, optimizing student recruitment and institutional placement has become a crucial challenge for vocational higher education. This study proposes an integrated analytical framework that combines Random Forest prediction with Propensity Score Matching (PSM) causal inference to evaluate the impact of algorithmic student– institution matching on enrollment quality, academic achievement, and retention outcomes. Using three-year empirical data (2021– 2023) from a vocational college in southern China comprising 6,200 anonymized records, the Random Forest model achieved high predictive performance (AUC = 0.87) and generated a continuous matching score representing the compatibility between students and institutions. The PSM analysis revealed that students with high matching scores exhibited a 12.5% higher enrollment rate, a 0.28-point increase in GPA, and a 6.9% reduction in dropout risk compared with their matched peers. Robustness tests confirmed the stability of these effects across alternative matching specifications and subgroups. The findings demonstrate that integrating machine learning and causal inference not only enhances the accuracy of admission predictions but also uncovers the genuine causal mechanisms underlying student success. This framework provides a replicable and interpretable methodological foundation for data-driven enrollment strategies, contributing to both the theory of student–institution fit and the practice of intelligent educational governance.
Author: Xiaosong Li
Institution: Guangzhou Nanyang Polytechnic College
Language: English
Subfield: Computer Science Applications
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7170177836