Exploring the Impact of Code Style in Identifying Good Programmers

Code style reflects the choice of textual representation of source code. This study, for the first time, explores whether code style can be used to identify good programmers with a vision that recruitment process in the software industry can be improved. For analysis, solutions from Google Code Jam were selected. The study used cluster analysis to find association between good programmers and style clusters. Furthermore, supervised machine learning models were trained with stylistic features to predict good programmers. Results reveal that, although association between programmers with particular clusters could not be concluded, supervised learning models can predict good programmers.