Creating a scalable predictive learning analytics model in line with instructional design practices

• In Event: Evaluating the Impact of Instruction

Short Description

The session will summarize a study that tested the viability of a model for predicting student performance in an undergraduate biology class. Instead of manual coding, this study utilized natural language processing techniques to classify question topics. It was found that this model was moderately viable, and may be appropriate for limited interventions. This session will include discussion of the value of such predictive models to practitioners as well as potential next steps in research.

Authors

• Presenter: Tanner Matthew Phillips, Indiana University

• Presenter: Gamze OZOGUL, Indiana University

• Contributor: Jamie Jensen, Brigham Young University