The Effect of Programming Language in Software Bug Prediction

Software maintenance is a very important phase in the life cycle of software development. As part of maintenance, we need to identify bugs within the code and fix them for every release. Software bug prediction (SBP) allows us to identify modules in the software that may have the tendency to be buggy. This enables us to perform targeted testing and properly plan maintenance cycles. In most research performed, we observed that dataset of software programs written in C language were used. Each programming language are inherently different and has it's own constructs, practices and nuances. In this research we focus on identifying if there exists a specific classifier that works well for each programming language which in turn is used to analyze the effect of programming language on Software Bug Prediction. Datasets of software written in C, C++ and Java has been collected and the most accurate classifier for each dataset of a programming language has been identified. The ML models used in this paper include Naive Bayes, Decision Tree and Random Forest.