The analysis of matching learners in pair programming using K-means

Programming is one of the educational fields that people of digital era have been taking a particular interest in. However, there has still been a shortage of programmer in the labor market as the majority of the graduates were relatively under par. This can be solved by accelerating the educational development to ensure learners are equipped with better quality. For the addressed problem, this study developed a matching system for pair programming. Pair programming theory believes that when an expert is paired with a beginner, it accelerates the beginner to progress more efficiently as oppose to coding alone. Nonetheless, the theory did not address the issue of programmer behavior, which is another important aspect in programming. Therefore, the study additionally employed k-means clustering to create a new cluster of programmers based on their common behaviors. This involved variables like programming competency (represented by A), learning behavior (represented by B), and behavioral interoperability (represented by C). The study employed a questionnaire based on a basic programmer test, the Learning and Study Strategies Inventory theory, and the Seven Synergistic Behaviors of Pair Programming theory. The analyzed data comprised 100 students from the Department of Computer Science. Grouping was done in four of the following categories (with sum square error of 8.92): 1)Those with high scores of A, B and C; 2) Those with high A scores, but with low B and C scores; 3) Those with low A scores but with high B and C scores; and 4) Those with low scores of A, B and C. Matching was conducted involving 10 matching patterns and the match with the fastest development of programming competency were group 1 matched with group 3. To reveal the pair analysis, the derived results were used to developed an online matching algorithm so that programming students could optimally pair with better learning speed.