An Interpretable Framework for Identifying At-Risk Students (Poster 6)

• In Event: AERA Poster Session Eleven
  In Poster Session: Models for Learning

Abstract

The relatively low learning achievements and high dropout rate in online learning contexts have been the major concerns of educators. This paper introduces a novel approach to identify at-risk students through analyzing learning activities at a finer granularity on a weekly basis. Specifically, students weekly learning associations were considered through converting the learning in previous week into meaningful possibilities to infer the following week’s performance. To make the predicted output more interpretable in an individualized manner, the demographic information was explored in the built model while predicting at-risk students. The experimental results demonstrated the good performance of the proposed model in comparison with the baseline models when identifying at-risk students both on the weekly and over the course measurements.

Authors

• Bo Pei, University of South Florida
  Presenting Author

• Wanli Xing, University of Florida
  Presenting Author