基于相似度的软件工程中软件成本估算问题研究  

Software Cost Estimation in Software Engineering Based on Similarity

作　　者：王楠 余建坤[1] WANG Nan, YU Jian-Kun(School of Information, Yunnan University of Finance and Economics, Kunming 650021, Chin)

机构地区：[1]云南财经大学信息学院,昆明650021

出　　处：《计算机系统应用》2018年第4期254-258,共5页Computer Systems & Applications

基　　金：云南省高校商务智能科技创新团队基金(42212217010)

摘　　要：软件工程概念从1968年被提出以来,经历了近50年的发展,软件系统规模和复杂程度日益加大,然而从上个世纪70年代左右开始,软件工程领域出现大量软件项目进度延期、预算超支和质量缺陷为典型特征的软件危机.这体现出软件成本估算在软件工程开发过程的重要性.精准的软件成本估算是软件工程按时完成的保证.本文采用一种基于皮尔逊相关系数的相似度量方法,结合TOPSIS方法软件成本进行类比估算以获取与之最接近项目的项目作为参考进行软件成本估算.最后将该方法应用于Desharnais数据集进行实验,并和其他方法进行比较,实验结果表明,本文采用的基于相关系数的软件成本度量方法较已有的相似性度量方法准确度更好.Since the introduction of software engineering concept in 1968, it has experienced the development of nearly 50 years. The scale and complexity of software system have been increasing day by day. However, since the 1970s, there has been a delay in the progress of a large number of software projects. And quality defects for the typical characteristics of the software crisis, is still frequent. Software cost estimation in the software engineering development process has been playing an important role. The precise software cost estimate is the guarantee of software engineering on time. In this study, a similarity measure based on Pearson correlation coefficient is used, and the software cost estimation is carried out by using the TOPSIS method to estimate the cost of the project with the closest project. Finally, the method is applied to the Desharnais dataset and compared with other methods. The experimental results show that the software cost measurement method based on the correlation coefficient has better accuracy than the existin~ similaritv measure mathnd

关 键 词：相关系数 成本估算 相似性度量 

分 类 号：TP311.5[自动化与计算机技术—计算机软件与理论]