基于缓冲的软件项目工作量估算模型研究  

Estimation Model of Software Project Effort Based on Project Buffer

作　　者：张俊光[1] 沈佳佳[1] 杨双[1] 

机构地区：[1]北京科技大学东凌经济管理学院,北京100083

出　　处：《管理工程学报》2015年第4期171-177,共7页Journal of Industrial Engineering and Engineering Management

基　　金：教育部人文社科基金资助项目(10YJC630384);中央高校基本科研业务费项目(FRF-BR-12-020)

摘　　要：为提高软件项目工作量估算模型的实用性,加强对软件项目工作量风险的管理,研究了基于历史数据的软件项目工作量估算模型及工作量缓冲计算模型。采用基于规模的分段函数进行软件工作量的估算,经拟合效果对比分析发现,运用分段函数进行工作量估计可以有效地提高估算的精确度;引入MER和MRE对离散型的数据进行拟合,用三条曲线方程进行工作量估计,并通过计算欧氏距离来分别对函数赋予权重,使得估算结果更加精确;将进度缓冲的概念应用于软件项目工作量管理领域,提出了基于因素驱动分析的工作量缓冲计算模型。最后,通过一个算例,将所建模型的应用效果与实际项目进行对比分析,结果表明该模型能够有效地提高软件项目工作量管理的成功率和效率。Inaccurate effort estimation has serious impact on the rationality of software schedule planning. Experts have been working to improve software effort estimation accuracy, and have proposed many software effort estimation methods in recent years. However, the research result indicates that these methods still lack practicality.In addition, adding buffers to prevent project risks has become an important method in the field of project schedule management. Some surveys revealed that buffer management has substantially improved projects＇ timely completion rate. However, there are still many projects that cannot be completed on time, and there are only few buffer studies on software projects. Moreover, current research mainly focuses on project schedule buffer; there is little software effort buffer research. Accordingly, to improve the practicability of software project effort estimation models and strengthen the risk management of software project effort, a software project estimation model based on historical data and an effort buffer model was proposed. Firstly, piecewise function based on software size was used to estimate software project effort. Comparison analysis indicates that this method could significantly improve the estimation accuracy. In addition, MER and MRE were introduced to conduct discrete data fitting. Three curve equations were adopted for effort estimation, and the Euclidean distance was calculated to give the functions weight respectively. The improved estimation method has three main advantages. 1） It can estimate the effort according to historical data without over-reliance on the experts＇ experience. 2） The data is divided into two kinds, small-scale and large-scale, in accordance with the size; and the project effort is calculated with the piecewise function by different formulas. Considering the relationship between project size and effort is not a simple linear relationship, especially when the size reaches a certain extent. The proposed method is more reasonable than traditional

关 键 词：软件工作量估算 工作量缓冲 分段函数 历史数据聚类 欧氏距离优化 

分 类 号：TP311.5[自动化与计算机技术—计算机软件与理论]