Code Complexity Analyser and Visualiser for Novice Programmer

Code complexity can have a significant influence on software quality. With studies showing program developed by novice programmers can influence software complexity due to lack of experience, practice, and understanding of the concept of programming. This paper investigates the utilisation of machine learning techniques to analyse code complexity levels. Using a public collection of JavaScript codes, we developed a machine learning model to identify the relationship between code characteristics and complexity level. We selected six methods and performed k-fold cross-validation. It was observed that Classification and Regression Trees (CART) and K-Nearest Neighbours (KNN) yielded the best prediction results. Finally, we also implemented a visualisation tool to present the code analysis results providing a means to gain insights on JavaScript codes through their characteristics and complexity level.