Year: 2025
Type: article
Abstract: Using stacked ensemble machine learning, we aim to predict the dropout of pre-service teachers. By training the data of 2661 students, we investigate the extent to which admission test data and demographic factors predict pre-service teachers’ dropout and how the integration of student success data can improve the accuracy of these predictions. We utilise an ensemble-model approach, consisting of three base models along with a meta-model, incorporating base model predictions to learn the strengths and weaknesses of base models, and therefore leveraging overall predictive performance. The results showed that 61% of students could be correctly classified (drop out/retention) using only admission test data and demographic information. When both admission test data and GPA were used, 76% of students were correctly classified. The model shows strong potential but also illustrates the need to develop more refined models that include historical data for early identification and intervention of at-risk students.
Institution: University of Graz
Language: English
Subfield: Computer Science Applications
Field: Computer Science
Domain: Physical Sciences
SDG: Quality education
Open Access status: hybrid
OpenAlex ID: W4412684259