Abstract:
This research to practice full paper provides preliminary evidence that integrating reflections is a significant feature to identify at-risk students early in a semester ...Show MoreMetadata
Abstract:
This research to practice full paper provides preliminary evidence that integrating reflections is a significant feature to identify at-risk students early in a semester as verified and validated by a sequence-based learning analytics model. We've devised an active learning classroom model which incorporates reflection at multiple points in students' learning experience. This active learning model adopts Kolb's Learning model to provide a coherent and connected set of activities before, during, and after the class. Unlike periodic assessment through testing, reflections can provide nearly-real-time information about student's experiences in class. We extract sentiment feature vectors to capture students' affect from written reflections. These features typically aren't assessed on tests or during in-class activities. These features were extracted automatically using LIWC (Linguistic Inquiry and Word Count) is a tool for applied natural language processing) which is less cumbersome to implement than manually reading the written reflections. We find that using these sentiment feature vectors extracted from the reflections in our learning model increased accuracy while decreasing time-to-detect at-risk students significantly.
Published in: 2018 IEEE Frontiers in Education Conference (FIE)
Date of Conference: 03-06 October 2018
Date Added to IEEE Xplore: 07 March 2019
ISBN Information: