- Title
- An architecture for feedback-driven learning analytics
- Creator
- Winfield, Philip John
- Subject
- Educational statistics -- Data processing
- Subject
- Education -- Research -- Statistical methods
- Subject
- Teacher-student relationships
- Date Issued
- 2023-12
- Date
- 2023-12
- Type
- Master's theses
- Type
- text
- Identifier
- http://hdl.handle.net/10948/62744
- Identifier
- vital:72935
- Description
- Feedback from students provides an opportunity to gain insights into students’ learning behaviour and participation in higher education learning and teaching. Academic staff are expected to constantly review and improve the learning and teaching environment where feedback contributes vital information toward pedagogical decision-making. Reviewing student feedback is essential but time-consuming, making it crucial to explore more effective and efficient ways to analyse and present feedback that encourages intervention and support. The general design science research evaluation pattern guided the construction and evaluation of a Feedback-Driven Architecture (FDA) for Learning Analytics (LA) to address this problem. An FDA implementation using a suitable case demonstrated each component validating the feasibility and effectiveness of the proposed design. The components identified for inclusion in the FDA were integrated within layers of a three-tiered architecture pattern. The data layer incorporates the collection, preparation and storage of learning management system data which includes free-form narrative student feedback. The application layer contains logic to support the analysis of free-form narrative student feedback and extraction of learning analytics. Arrangement and visualisation of information in the presentation layer aim to promote appropriate intervention and support. Evaluation using a case study showed that the FDA provided necessary guidelines for developing an implementation that produced feedbackdriven learning analytics. Data preparation transformed unstructured data into a suitable representation for effective analysis. Cohorts within narrative feedback responses were identified using the k-means clustering algorithm and latent Dirichlet allocation. Logistic regression and support vector machines were applied as predictive models and trained using extracted quantitative markers to predict academic success.
- Description
- Thesis (MSc) -- Faculty of Science, School of Computer Science, Mathematics, Physics and Statistics, 2023
- Format
- computer
- Format
- online resource
- Format
- application/pdf
- Format
- 1 online resource (iv, 160 pages)
- Format
- Publisher
- Nelson Mandela University
- Publisher
- Faculty of Science
- Language
- English
- Rights
- Nelson Mandela University
- Rights
- All Rights Reserved
- Rights
- Open Access
- Hits: 556
- Visitors: 567
- Downloads: 13
Collections
NMU School of Computer Science, Mathematics, Physics and Statistics
| Thumbnail | File | Description | Size | Format | |||
|---|---|---|---|---|---|---|---|
| View Details Download | SOURCE1 | Winfield, PJ.pdf | 5 MB | Adobe Acrobat PDF | View Details Download |