Using Fine Grained Programming Error Data to Enhance CS1 Pedagogy

The paper reports on our experience using the log files from Spinoza, an online IDE for Java and Python, to enhance the pedagogy in Introductory Programming classes (CS1). Spinoza provides a web-based IDE that offers programming problems with automatic unit-testing. Students get immediate feedback and can resubmit until they get a correct program or give up. Spinoza stores all of their attempts and provides orchestration tools for the instructor to monitor student programming performance in real-time. These log files can be used to introduce a wide variety of effective pedagogical practices into CS1 and this paper provides several examples. One of the simplest is forming recitation groups based on features of student's problem solving behavior over the previous week. There are many real-time applications of the log data in which the most common errors that students make are detected during an in-class programming exercise and those errors are then used to either provide debugging practice or to provide the examples of buggy programming style. Finally, we discuss the possible use of machine learning clustering algorithms in recitation group formation.