Krambeer, Malte Leonard (2024) Identification of Prior Knowledge for the Optimization of Adaptive Fact-Learning. Bachelor thesis, Psychology.
|
Text
Krambeer2024Identification-of-Prior-Knowledge-for-the-Optimization-of-Adaptive-Fact-Learning.pdf Download (330kB) | Preview |
Abstract
The present study demonstrates the possibility of predicting initial item knowledge for English vocabulary and Geography facts using learning data from the user’s interaction with the MemoryLab adaptive learning system. Considering accuracy, speed of forgetting, and response time as potential predictors of initial item knowledge, we evaluate inclusion rates and regression coefficients of lasso regression models fitted during a k-fold cross-validation procedure. Additional modified prediction models are examined for the potential of adaptation to specific learning situations. We conclude that user and item accuracy measures provide the highest predictive power, with measures such as speed of forgetting and response time providing limited and more domain-specific value. Preliminary findings imply the potential to optimize adaptive fact learning by removing predicted known items from the learning set, and should be confirmed by replications in a realistic setting.
| Item Type: | Thesis (Bachelor) |
|---|---|
| Supervisor name: | Rijn, D.H. van |
| Degree programme: | Psychology |
| Differentiation route: | None [Bachelor Psychology] |
| Date Deposited: | 30 Aug 2024 14:45 |
| Last Modified: | 30 Aug 2024 14:45 |
| URI: | http://gmwpublic.studenttheses.ub.rug.nl/id/eprint/4342 |
Actions (login required)
 |
View Item |