Investigation on of metric based software bug prediction approaches and a new software bug prediction recommendation

The aim of the software project is not only to produce a product that works correctly. It is also necessary to evaluate the quality of the software performed and to measure the quality. The more quality and error-free the software, the lower the costs during the maintenance process. One of the important points affecting the quality of the software is the number of bugs in the software. For this reason, it is very important to detect bugs in the software developed at an early stage. In the scope of the study, source code metrics which is one of the metric based software bug estimation approaches were examined. In the scope of the study, source code metrics from metric based software bug prediction approaches were examined and the relationship between software bug numbers and software quality metrics was analyzed. For this purpose, 25 open source game projects developed with java programming languages were used as data sets. "Understand" static code analysis tool was used to analyze software quality metrics. In the study, Spotbugs bug detection tool was used to determine the bug numbers of the projects. Linear regression method was used to determine the relationship between software bugs and software quality metrics. The prediction accuracy of the results obtained from the analysis results was made with leave one out cross validation. The results show that it is possible to make use of software quality metrics to estimate the number of software bugs.