Predicting expertise among novice programmers with prior knowledge on programming tasks

The studies on program comprehension have seen developments over the years from the cognitive science perspective. As eye-tracking technology has proven to analyze visual attention and gaze-performance, it has then been largely used in the program comprehension studies to help understand the underlying cognitive processes among the participants. In this research work, we conducted an experiment using common fundamental programming questions on 66 undergraduate computer science students to study the gaze-behavior among the high and low-performing participants on programming comprehension. We aim to better understand the differences in the time taken by the individuals in terms of their performance with existing prior knowledge and use machine learning to predict their expertise. Findings from this study suggest that mental schemas do play a role as the high performers demonstrated less time taken to attempt the questions than the low performers and machine learning algorithms were able to successfully predict their expertise. The conclusions drawn are supported by eye-tracking metrics across individual- and group-levels.