Answers are grounded in this paper and cite their sources.
Identifying students with low academic performance can help educational institutions determine when further review and support may be needed. This study evaluates whether assessment-activity features can classify students according to their mean assessment score using a Decision Tree algorithm. The data were obtained from studentAssessment.csv in the Open University Learning Analytics Dataset (OULAD), comprising 173,912 initial records. After 173 records with missing assessment scores were removed, the analysis included 173,739 valid records from 23,351 students. Low academic performance was operationally defined as a mean assessment score below 50. The data were divided using stratified sampling into 80% training and 20% testing sets, with additional evaluation using Stratified 5-Fold Cross-Validation on the training data. On the test set, the model achieved an accuracy of 68.83%, precision of 16.86%, recall of 75.48%, F1-score of 27.56%, and ROC-AUC of 79.59%. The number of valid assessment records per student had the highest feature importance at 59.75%. However, among the 1,643 students predicted as having low academic performance, 1,366 were false positives. Furthermore, the activity features were calculated using the entire observation period. Therefore, this experiment represents retrospective classification rather than prospective early-warning prediction before all assessments are completed. The model may consequently be considered only as a screening aid that requires further review. Temporal validation, comparisons with baseline models, and classification-threshold evaluation remain necessary.
No takes yet. Share an insight, caveat, or question.
Irmawati et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: