Data Mining of Syntax Errors in a Large-Scale Online Python Course

This paper investigates the common syntax errors students encounter when programming in Python using data from a large-scale online beginner course. We firstly analyse the error distribution to find differences between passing and failing students and then use clustering and Markov chains to identify clusters of student submissions with similar error pattern and how students move between these clusters during the course and between consecutive tasks. This type of analysis can be used by educators to understand student behaviour related to syntax errors and provide effective teaching support.