一种基于p-值修正法的软件错误数估算的可靠算法

A Reliable Algorithm for Software Error Estimation Based on Revision Method of p-valule

软件内部错误数估算是降低软件工程风险、保证软件工程顺利实施的重要途径。在当前软件测试中,项目管理人员采用传统的估算策略,但是估算效果往往受制于测试人员独立性等诸多因素的干扰。其中,具有相似背景的测试人员对于测试过程会具有相似认知,这种不独立性使软件错误数的模型估算值往往比真实值偏低。类似潜在因素在软件错误数估算中大量存在,降低了估算准确性。因此,本文基于传统的Hyman估算法提出了一种改进的度量方法(p-值修正法),该方法能有效排除组间人员相关性对Hyman模型的干扰,能在很大程度上提高估算的准确性。同时,该方法易于用户理解、简便易行、可靠性高,可降低软件工程风险,对决策人员有很大帮助,适合普遍推广。

Estimating the inner error number of software is an important means of decreasing software engineering risks and making software engineering implemented successfully.In current software testing,project managers employ traditional estimation strategy,and the estimation effect is affected by the factors such as the independence of the test personnel.Testers of similar background may have similar cognition of the testing process,and such independence makes the estimated number of errors smaller than the real number.Moreover,such potential factors exist a lot in error estimating, hence destroying the estimation accuracy.Therefore,an improved estimation method,p-value revision method,is proposed in this paper,which is able to remove the intervention by the inter-group correlations of the testers,promoting the estimation accuracy.Meanwhile,this method is easy to be understood by users,is ready to be implemented,and is highly reliable.It is capable of decreasing software engineering risks,is helpful for decision makers,and is suitable for wide use.