XAI기법을 활용한 대학생 중도탈락 주요요인 탐색 및 예측 모델 구축

초록

The dropout rate among university students in South Korea has been steadily increasing in recent years. Since dropout rates are now utilized as key performance indicators in evaluating domestic universities, institutions are actively conducting various research studies to mitigate this issue. One of the major strategies involves the development of predictive models that can identify at-risk students early through early warning systems, thereby facilitating proactive interventions to reduce dropout rates. In 2024, the average university dropout rate in Korea was reported at 5.73%, reflecting a 0.4 percentage point increase from 2023. High dropout rates have significant negative consequences not only for students and faculty but also for universities' reputations and national competitiveness. while entailing substantial socio-economic costs. Therefore, early identification and proactive support for students at risk of dropping out is of paramount importance. In recent years, with the advancement of artificial intelligence and data analysis techniques, interest in predicting student dropout using machine learning has significantly increased. Numerous studies have been published in prestigious international AI journals, identifying various factors that influence student dropout. This study aims to develop and compare various machine learning-based models for predicting student dropout. Furthermore, it incorporates explainable artificial intelligence (XAI) techniques, specifically SHAP and LIME, to interpret the influence of individual variables on the prediction outcomes and present methods for their utilization.

키워드