???????????????????????????????????????????? ????????? 2 ?????????????? ????????????????????????

Abstract

Currently, student dropout rates in Thailand have become a significant issue in every higher education institution in the country. The increasing trend of student dropouts continuously affects the revenue of each faculty and the overall image of the university. In this research, a predictive model for student dropout was developed. Factors influencing student dropout at the undergraduate level in the 2nd Year of the Faculty of Science, Prince of Songkla University, Hat Yai Campus, from the academic years 2018-2022, involving 3,092 students, were analyzed using machine learning techniques, specifically Classification. Features were selected to reduce the dimensionality of the data, and algorithms including Support Vector Machine with different kernels such as Linear, Polynomial, Radial Basis Function Kernel (RBF), and Sigmoid, Logistic Regression Algorithm, and Random Forest Algorithm were compared. The performance of the models was tested using 10-fold cross-validation and compared between training-testing data ratios of 70:30 and 80:20. It was found that the performance of predicting dropout rates with a 70:30 data split ratio was better than with an 80:20 ratio. The most suitable and time-efficient dropout prediction model selected had only 5 features, consisting of grades from 4 fundamental science courses and the major, with data augmentation using the SMOTE technique and employing the Logistic Regression Algorithm. This model achieved an accuracy of 91.64%.

????????????????????????????????????????????????????????????????????????????????????????????????????? ?????????????????????????????????????????????????????????????????????????????????????????????????????????????????? ?????????????????????????????????????????????????????????????????? ??????????????????????????????????????????????????????????????????????????????????????? ????????? 2 ?????????????? ???????????????????????? ??????????????? ????????????????? 2561-2565 ????? 3,092 ?? ???????????????????????????????????????????????????????????????????????????????????????????????? ???????????????????? ???????????????????????????????????????????????????????????????????????????????? ?????? Linear, Polynomial, Radial Basis Function Kernel (RBF) ??? Sigmoid ?????????????????????????????????????????????? ??????????????????????????????????????????????????????????????????????? SMOTE ?????????????????????????????????????????????????????????????? 10 ???? ?????????????????????????????????????????????????????????????????????????????????????????????????????? 70:30 ??? 80:20 ?????????????????????????????????????????? ???????????????? ??????????????????????????????????????????????????????????????????????????? 70:30 ?????????????????? 80:20 ??????????????????????????????????????????????????????????????????????????????????? ??? ????????????????????????????????????????????? 5 ????????? ?????????? ??????????????????????????????????? 4 ??????? ??????????? ????????????????????????????? SMOTE ???????????????????????????????? ???????????????????????????????? ?????? 91.64