摘要
针对软件项目前期成本估算不准确问题,通过构建软件项目案例库,提出一种基于CBR的软件项目成本估算方法(CBRCEM)。根据COCOMO模型成本驱动因子理论,确定影响项目成本属性特征;引入归一化效用函数,应用层次分析法计算影响项目成本属性的权重值;通过对常用案例检索算法的比较分析,结合软件成本估算的特性,建立基于改进的灰色关联分析理论的软件项目相似度算法;根据PERT理论估算软件项目成本,使估算结果更为准确。CBRCEM在Windows平台上用JAVA语言开发完成并在实际中加以应用,案例研究结果表明,对于软件项目前期成本估算,该方法能够得到更加准确的评估结果。
The software pre-project cost estimation is not accurate. To solve the problem, a software projects case library was built and a software project management cost estimation method on the case-based reasoning （CBRCEM） was presented. Accor- ding to the theory of cost driving factors in the COCOMO model, the attribute characteristics affecting project costs were deter- mined. The normalization utility function was introduced, and attribute weights affecting project costs were calculated using AHP. Through the comparative analysis of common algorithms for the case retrieval, combining with the characteristics of software cost estimates, the software project similarity algorithm based on the improved gray relation analysis theory was established. According to the PERT theory, software project costs were estimated that were more accurate. CBRCEM was developed and completed using the JAVA language on the Windows platform, then applied in practice. Case studies show that the method can get accurate assessments for the software preproject cost estimation.
出处
《计算机工程与设计》
CSCD
北大核心
2014年第11期3837-3844,3902,共9页
Computer Engineering and Design
基金
国家自然科学基金项目(70971124)
住房和城乡建设部科技计划基金项目(2010-R3-13)
烟台市科技发展计划基金项目(2012GGA000186)
关键词
软件项目
案例推理
灰色关联理论
成本估算
估算方法
software project
case-based reasoning
gray relational theory
cost estimation
estimation method
