软件成本估算及其不确定性研究
文献类型：学位论文
| 作者 | 杨达 |
| 学位类别 | 博士 |
| 答辩日期 | 2009-01-12 |
| 授予单位 | 中国科学院软件研究所 |
| 授予地点 | 软件研究所 |
| 关键词 | 软件成本估算 估算模型 不确定性 涉众协作 多赢协商 |
| 其他题名 | Research on Software Cost Estimation and Its Uncertainty |
| 中文摘要 | 软件成本估算作为软件项目可行性分析、预算、计划以及控制的基础，是软件工程中的一个重要研究领域。自上世纪60年代以来，尽管软件成本估算一直受到研究者的持续关注，但在现实环境中软件成本估算仍然是软件行业面临的一项难题，仍然有很大的深入研究与改进空间。 现实环境下，软件成本估算方法需要接受不完整且不能完全确定的信息，估算可能的软件开发工作量与开发周期，并衡量估算结果的不确定性与风险。软件成本估算方法还需要满足为使用者创造价值、实施成本低、得到人文与技术上支撑等前提条件，才能在现实环境中被接受。成本估算的应用，也还需要随着项目的演进，与涉众协商、项目计划与项目监控过程密切互动。不能处理好现实环境中的不确定性，不能解决方法实施和应用中所面临的多种关键困难，是已提出的大量成本估算方法和模型难以在现实环境得到广泛使用并发挥影响的重要原因。 本文从确定问题、方法改进、方法应用、工具支撑等多个角度，对软件成本估算进行了较为系统的研究，为解决在现实环境改进成本估算现状所面临的关键困难，以及处理软件成本估算的不确定性这个核心问题，提出了一套包含方法、过程与支撑工具的比较系统、完整的针对实际需要的解决方案。本文研究工作的主要贡献包括： 1）提出软件成本估算的问题模型。设计并实施了中国软件行业的软件成本估算现状调查，探索了我国软件成本估算的现存问题，以及改进成本估算所面临的困难。结合文献综述与产业调查结果，运用“技术接受与使用模型”以及“结果链”等方法，提出软件成本估算的问题模型。涵盖技术、人文、经济与管理多个角度，系统地总结了软件成本估算所面临的问题和潜在的改进。 2）提出集成的成本估算方法。不同于目前估算方法均依赖某种固定估算模型，而是将多种子估算模型作为有用的信息输入，在不同环境下基于历史项目数据自动生成适应该环境的集成的估算模型。 3）针对成本估算在应用环节所面临的关键困难，提出了WikiWinWin软件项目多赢协商方法并开发相应的支持工具。帮助项目涉众正确认识并使用成本估算，促进项目演进过程中涉众协商、成本估算及项目计划与执行过程的有效融合，使成本估算更有效发挥作用。 4）提出了分析框架与相应的综合性方法以处理软件成本估算的不确定性。对软件成本估算的不确定性这个核心问题，进行了系统的分析。使用贝叶斯网络以及Monte Carlo仿真对现有估算模型进行扩展，以处理估算输入的不确定性；使用集成成本估算解决估算模型本身的不确定性；并在成本估算的应用环节，以WikiWinWin方法为核心处理软件成本估算的不确定性。 5）设计开发了软件成本建模与估算支撑工具。结合本文前面提出的成本估算方法形成了工具支撑下的集成的成本建模与估算方法（InCoME方法），在处理估算的不确定性、准确性、稳定性、客观与可重复性、透明性以及建模与估算的自动化支持方面都具有较好的能力，较为全面地满足了企业在现实环境中的需要。 |
| 英文摘要 | Software cost estimation is the basis of project feasibility analysis, budgeting, planning, and control. It has been an important and active research area of software engineering for more than 40 years. However, software cost estimation is still a challenging issue for the software industry, and it still requires more intensive research and improvement. In a real project, software cost estimation has to cope with the incomplete and uncertain information until the end of project lifecycle, estimate the effort and schedule, and assess the risk of effort or schedule overrun. To be accepted and used in industry, software cost estimation methods need to fulfill many prerequisite such as low adoption cost, creating sufficient value for users, provided with social and technical support, etc. Besides, software cost estimation has to interact with stakeholder negotiation, project planning and control as projects evolving. A lot of existing cost estimation methods cannot properly solve the critical problems in real projects, so they are seldom used in the industry. This thesis addresses the challenges of software cost estimation by working on problem analysis, methods improvement, methods application, tool support, etc. And the thesis provides a relatively systematic and complete set of methods, process, and tools to resolve the critical software cost estimation issues. The main contributions of this thesis include: 1）Proposed a problem model for software cost estimation. The author conducted a survey on software cost estimation in the Chinese software industry, and explored the current problems of software cost estimation and the barriers and difficulties of further improvement. Then, the author systematically analyzed the software cost estimation problems and potential improvements by using the "technology acceptace and usage model" and the "benefit realization chain" methods. The proposed problem model of software cost estimation covered various problems and potential improvements in the technical, social, economic, and management aspects. 2）Proposed an integrated software cost estimation method. The new method is different from the current formal cost estimation methods, because it does not rely on a specific cost estimation model, but use several different cost estimation models as useful information to generate appropriate new cost estimation model according to the environment. 3）Proposed the WikiWinWin methods and developed related tools to resolve the critical issues of conducting cost estimations as the project evolving. The WikiWinWin methods and tools can help the project stakeholders to correctly recognize and use cost estimations. It can also make the cost estimation more efficient by improving the interactions among cost estimations, stakeholder negotiation, project planning and control. 4）Proposed a framework to analyze the uncertainty of software cost estimation and the integrated methods to resolve the uncertainty. To cope with the uncertainty of the estimation input information, the author used the Bayesian belief networks and the Monte Carlo simulation to extend the existing cost estimation models. To cope with the uncertainty of the estimation methods, the authors used the integrated cost estimation method. Beyond estimation, the author used the WikiWinWin methods to organize several critical practices to cope with the uncertainty. 5）Developed the integrated software cost modeling and estimation support tool. The new cost estimation methods proposed in this thesis with the tool support, can be called the InCoME method (Integrated Cost Modeling and Estimation). The InCoME method has strengths in accuracy, stability, objectivity, repeatability, transparency, automation of modeling and estimation, and resolving uncertainty, so it is strong at satisfying the needs of software companies in real-life situations. |
| 语种 | 中文 |
| 公开日期 | 2011-03-17 |
| 页码 | 153 |
| 源URL | [http://124.16.136.157/handle/311060/6868]  |
| 专题 | 软件研究所_基础软件国家工程研究中心_学位论文 |
| 推荐引用方式 GB/T 7714 | 杨达. 软件成本估算及其不确定性研究[D]. 软件研究所. 中国科学院软件研究所. 2009. |
入库方式： OAI收割
来源：软件研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。