대학생 취ㆍ창업 지원 프로그램 참여 예측 요인 탐색

In response to the growing instability of youth employment, universities have been expanding career and entrepreneurship support programs. However, persistently low participation rates hinder the effectiveness of these initiatives. This study empirically examines the key factors predicting university students’ participation in career and entrepreneurship support programs by applying machine learning techniques. In particular, the study distinguishes between participation in curricular (“for-credit”) programs and extracurricular (“non-credit”) programs to identify predictive factors unique to each type. Using data from the KEEPⅡ survey (1st-8th waves), the analyses employed XGBoost for prediction modeling and SHAP values for model interpretation. The results indicate that participation in curricular programs was most strongly predicted by factors such as PBL course experience, self-concept, GPA, time spent studying alone, and experience with video-based lectures. For extracurricular programs, major predictive variables included the number of club activities, engagement in foreign language study, PBL course experience, total scholarship amount received, and capstone course experience. Based on these findings, the study suggests that universities should enhance program participation and effectiveness by designing integrated programs aligned with innovative instructional methods such as PBL and capstone learning, improving program quality through systematic needs assessment, and strengthening information accessibility by leveraging students’ social networks.