University of Twente Student Theses
Comparing the Effectiveness of Ensemble Models and Single Algorithms in their Success in Predicting At-Risk Students
Wu, Weijun (2025) Comparing the Effectiveness of Ensemble Models and Single Algorithms in their Success in Predicting At-Risk Students.
|
PDF
1MB |
| Abstract: | Detecting at-risk students at an early stage is key to intervene in a timely manner and aid them to improve academic results. Learning Analytics, the process of analyzing the learning patterns and behaviour of students, paired together with Machine Learning or Deep Learning models, can help predict their expected final results. Ensemble models (the combination of multiple models) and single algorithms are often used as analytical techniques. This study proposes to do an analysis on the effectiveness of these two different strategies by using accuracy, precision, recall and F1-score. The experimental results concluded that ensemble models have a higher average F1 and recall score at the later stages of the courses. The stack generalization in particular, produced the best results with an F1 and recall score of 87.55% and 87.89% respectively. The highest achieved F1 score being produced by the Random Forest model at 88.03%. Whilst single algorithms outperformed the ensembles in earlier stages. These results can be used to help educators decide which technique to apply when determining a model with student learning data. |
| Item Type: | Essay (Bachelor) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 54 computer science |
| Programme: | Computer Science BSc (56964) |
| Link to this item: | https://purl.utwente.nl/essays/105125 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page