Why first-year e-students are dissatisfied: Machine learning methods for enhancing retention

Due to advances in wireless technology, e-learning programs and e-courses have increasingly been employed as a mainstream educational mechanism and potentially become crucial incomes of higher education (HE) worldwide. In practice, e-students are required to have highly qualified e-learning programs and satisfied services. It is also extremely difficult for HE to maintain e-students retention as e-students, especially first-year e-students, easily exit from their e-learning programs or shift from one HE to another HE owing to dissatisfaction. However, the dissatisfaction of first-year e-students has gained limited theoretical and practical attention. Thus, it is essential to explore what features make first-year e-students dissatisfied so that HE may have enough time to issue preventive strategies at the early stages for sustainable e-learning adoption. Thus, this study aimed to extract important features using machine learning methods. Data was obtained by using a 5-point Likert e-questionnaire between May and June 2022, generating 499 valid responses from first-year e-students in a Vietnamese public university. The results showed that DT (90.4%) was superior to SVM (88.8%), LR (88.8%), and MLP (85.0%). The most important features included“easy access e-courses via the school e-learning platform”,“adequate personal internet skills”,“feeling stimulated to attend e-courses,“stable and uninterrupted e-learning platform”,“adequate personal digital devices”,“teachers’great efforts to improve students’learning”, and“timely responses provisions to students’inquiries”. The findings of this study are expected to assist HE policy-makers in minimizing e-students’dissatisfaction and maximizing their satisfaction in order to enhance e-student recruitment and retention, and enhance the quality of e-educational programs.