Programming Trajectory Classification in Block-Based Learning Environments

• In Event: Computational Analyses of STEM+C Learning

Abstract

This study aims to classify the correctness of student solutions in block-based programming learning environments. The approach involves three steps: (1) transformation of successive program edits into an intermediate representation; (2) extraction and segmentation of information of different types and levels of granularity; and (3) evaluation of the quality of classification. Analysis was conducted with 444 distinct programming trajectories while students solved a problem on nested loops. The results showed that the best performing models were obtained when segmenting the edit sequences earlier during the learning session and in taking into account a greater number of edits. We discuss the implications for a hint delivery system that provides support to students facing difficulties during the early stages of problem solving.

Authors

• Punarva Vyas

• Eric G. Poitras, Dalhousie University

• Angela Siegel

• Raghav Sampangi