基于分类算法的软件缺陷原因分析

摘要

软件缺陷预防是软件质量保证的重要手段,而软件缺陷原因分析又是软件缺陷预防的核心任务.针对当前的缺陷原因分析方法对大型项目缺陷数据无法快速、深入地定位软件缺陷原因的问题,提出了一种基于机器学习分类算法的缺陷原因分析方法.通过对当前两种原因分析方法的比较,建立改进的缺陷量化方法；然后比较主流的分类算法,应用算法对量化数据分类.实验结果表明,该方法有较好的实用效果,同时极大的减少了分析代价.%Software defect prevention is an important matter in software quality assurance, in which software defect cause analysis is the core task. Aimed at the problem of the current methods, which could not be used to analyse defect cause quickly and go into the root of the defect in large projects, a method of defect cause anylysis based on machine learning classification algorithm is presented. An improved method of defect quantization is built through comparing the current two methods. Then common classification algorithms are compared, and the quantized data is classified. The experiment shows that the method presented has good effect and reduce the analysis cost greatly.