主成分分析法在基于度量的软件风险评估中的应用  

Principal Component Analysis' Application to the Software Metrics-based for Risk Assessment

作　　者：潘春光[1] 陈英武[1] 汪浩[1] 

机构地区：[1]国防科技大学信息系统与管理学院,湖南长沙410073

出　　处：《运筹与管理》2005年第5期80-84,共5页Operations Research and Management Science

基　　金：国家自然科学基金资助项目(70272002)

摘　　要：软件风险评估一般分为主观评估方法和客观评估方法两种。主观评估方法多依赖于专家知识及管理人员的经验,客观评估方法则是从软件产品本身的内在属性出发,通过对软件产品复杂性等特性的度量来进行的。本文研究了现代统计分析技术中的主成分分析法在基于度量的软件风险评估中的应用,并对面向对象开发的软件产品给出了算例,从而能帮助软件开发者或管理人员在项目管理过程中识别软件产品的高风险模块,便于有效地开展风险管理。Software risk assessment techniques can be divided into subjective methods and objective methods. The subjective methods are often dependent on the knowledge of experts and the experiment of managers, while the objective methods are based on the metrics of software products＇ attributes such as complexity. In this paper, the principal component analysis, which belongs to the modern statistic analytical techniques, is used to assess the software risk. A case study for using PCA to assess risk of the object-oriented software development is also given. In doing so, the software developers or management people can identify the software high-risk modules easily and perform risk management effectively.

关 键 词：主成分分析法 软件度量 风险评估 

分 类 号：N945.1[自然科学总论—系统科学]