標題: | 軟體發展流程管制圖之經濟-統計性設計 Economic Statistical Design of Control Charts for Software Development Process |
作者: | 陳怡君 Chen, Yi-Chun 唐麗英 洪瑞雲 Tong, Lee-Ing Horng, Ruey-Yun 工業工程與管理學系 |
關鍵字: | 經濟-統計性設計;軟體發展流程;軟體測試階段;資料包絡法;Economic statistical design of control chart;Software development process;Software test process;Data envelopment analysis |
公開日期: | 2010 |
摘要: | 目前政府正積極向台灣軟體組織推動能力成熟度整合模式(Capability Maturity Model Integration, CMMI),以改善台灣軟體組織在軟體開發流程上的品質,進而提升台灣軟體產業之競爭力。透過CMMI之認證,軟體組織可以證明其具有一定水準的軟體發展流程管理能力,並符合軟體組織國際合作的基本要求。然而,許多軟體組織在導入CMMI時常會遭遇困難,查其原因在於軟體組織不易制定量化之目標。中、外文獻針對軟體產業的特性,已發展出適用於軟體發展流程管制圖的經濟性設計(Economic Design of Control Charts),但經濟性設計在成本模式中僅考慮到低成本,而無法同時兼顧到管制圖之最小型一及型二誤差。因此,本研究的主要目的是建構一個適用於軟體發展流程之經濟-統計性設計模式,在最小化成本、最小化型一及型二誤差等多個目標下,利用資料包絡法(Data Envelopment Analysis, DEA),得到最佳管制圖的參數設定值。本研究最後利用台灣某軟體公司之軟體生命週期(Software Life Cycle)的測詴階段資料,驗證本研究所提出的軟體發展流程管制圖之經濟-統計性設計模式確實有效。 Capability Maturity Model Integration (CMMI) is actively promoted by the government to improve the software quality in software development process of software organizations in Taiwan. The management capabilities of a certain level of software development process can be proven through the certification of CMMI by software organizations and the basic requirements of international cooperation in the software organizations could also be fulfilled. Many software organizations confronted difficulties for quantifying their goals when they continuously enhanced to higher maturity level of CMMI. Statistical quality control (SPC) can be employed to monitor the software development process. Previous studies for economic design of control charts only focused on minimizing the cost of constructing control charts for software development process and consequently have poor statistical properties such as high type I or type II errors. In this study, we developed an economic statistical design of control charts for the software development process. Three main attributes by this study have been characterized; expected cost, average run length and detected power designed with the control chart parameters. Based on data envelopment analysis (DEA), a solution procedure is proposed to solve the multiple criteria decision-making problem. Finally, an application of solution procedure is presented. |
URI: | http://140.113.39.130/cdrfb3/record/nctu/#GT079833502 http://hdl.handle.net/11536/47849 |
Appears in Collections: | Thesis |