http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
기계학습 기반 대학원생 학업중단 예측 연구: K 대학을 중심으로
김태윤 ( Taeyoon Kim ),윤병연 ( Byungyeon Yun ) 한국교육정보미디어학회 2024 교육정보미디어연구 Vol.30 No.2
This study aims to predict graduate student dropout rates and explore the factors influencing academic discontinuation. Machine learning methods were utilized, specifically the XGBoost algorithm within the boosting family. Data from students admitted to K University from 2020 to 2022 were analyzed, including their academic records, scholarships, and research activities. After evaluating the performance, the SHAP library was used to identify features that impact dropout. The top 10 influential features were enrolled months, grades, LMS login counts, changes in academic status, number of semesters off, total research funding, portal login counts, admission semester, scholarship amounts, and age. Educational implications derived from the results include the following: First, institutions need to address and listen to difficulties graduate students face to prevent dropout. Second, a model should be implemented to detect early signs of dropout risk, possibly by enhancing database. Third, attempts should be taken to increase graduate student participation in research and improve access to scholarship information. The limitations of this study include a disappointing recall rate, the need to adjust the threshold, and the lack of a specific system. Based on these implications and limitations, this study underscores the necessity and method for developing a system to predict graduate student dropout, advocating for the acquisition and utilization of additional data for further research.