2013 Poster Sessions : Predicting Help-Seeking Behavior in Students Learning to Program

Student Name : Engin Bumbacher
Advisor : Paulo Blikstein
Research Areas: Artificial Intelligence
Abstract:
Recent research in CS education has leveraged machine learning techniques to extract characteristic pathways of students progressing through assignments in programming courses. These pathways show capabilities for predicting student performance in these classes. This suggests that such data could be used to quickly develop effective instructional strategies. With this in mind, we present a methodology for characterizing different program structures and their relationship with a student's help-seeking behavior, as expressed in data logged at teaching assistant facilities. Preliminary findings show that a hybrid set of non-semantic and semantic features of a student's code are predictive of whether or not a student got help. Based on these findings, we plan to develop a model of the impact of teacher intervention on a student's pathway through homework assignments.

Bio:
I am a first-year PhD student in Learning Science and Technology Design. My advisor is Prof. Paulo Blikstein at the Transformative Learning Technology Lab in the School of Education. I have a BSc in Physics and a MSc in Neural Systems and Computation, both from the Swiss Federal Institute of Technology, Zurich. Prior to graduate school, I co-founded Socos, a start-up involved in the development of adaptive learning technology for large-scale online courses.