Using Data Mining to Investigate the Effect of Cognitive Style on Programming Habits

Different programmers code in different ways. Knowing these habits and the human factors that affect them significantly impacts teaching and task assignments in programming. This article examines the effect of cognitive style on programming habits. We used a questionnaire to obtain data on cognitive style, programming experience, programming skill, interest in programming, and programming habits from 275 student programmers. After preprocessing and feature selection, we evaluated the effectiveness of different data mining techniques in estimating programming habits. Using the Support Vector Machine, the most effective method, we predicted each programming habit once without cognitive style and the second time with cognitive style. The results showed that cognitive style affects the programming habit of systematic debugging with Glass’s Delta value = 0.22. Programmers with a median score in cognitive style, both analytic and Intuitive, more often debug their codes systematically than programmers with lower or higher scores in cognitive style. Thus assigning programmers with both Intuitive and analytic talent would be more effective when projects need systematic debugging. Moreover, trainers should pay more attention to only Intuitive or only analytical students when teaching systematic debugging. We recommend teachers, trainers, and managers consider cognitive style in programming.