軟體發展流程管制圖之經濟性設計

Abstract

統計製程管理(Statistical Process Control; SPC)是工業界為了減少製程或產品之變異以及提升製程能力而使用之品質管制工具。管制圖(control chart)又是最常被使用之SPC工具，此工具是應用統計手法來判定造成製程變異的原因是機遇原因(chance causes)或非機遇原因(assignable causes)。管制圖也可以被應用在監控及改善軟體流程(software process improvement; SPI)上，特別是軟體發展流程(Software Development Process)中的軟體測試流程(Software Test Process; STP)。 在考量管制成本下，管制圖的樣本大小(n)、抽樣間隔時間(h)、管制界限寬度(k)等可利用管制圖之經濟性設計(economic control chart)來決定。本研究之主要目的是根據軟體產業的特性，修飾用於傳統製造業之管制圖經濟性設計手法，另提出一套適用於軟體發展流程之管制圖經濟性設計，以供軟體公司在特定之成本參數下，最佳化管制參數(n,h,k)，使每一單位時間之期望成本最小。本研究最後利用台灣某軟體公司之軟體生命週期 (Software Life Cycle)中的測試階段資料，驗證了本研究所提的軟體發展流程之測試階段品質管制成本模式確實有效。

Statistical Process Control (SPC) is widely utilized in industry to reduce the process/process variability or enhance the product quality. Control chart is one of the most frequently employed SPC tools. It can be used to distinguish the process variability is caused by chance cause or assignable cause based on statistical concepts and methods. Control chart can also be used for monitoring and improving the quality of the software process, especially the Software Test Process (STP) in Software Development Processes. To minimize the controlling cost, the economic design of the control charts can be utilized to determine the appropriate setting for the sample size(n), sampling frequency(h), and the control limits(k) of a control chart. Therefore, this study modifies the conventional economic design of control charts and develops the economic design for STP in Software Development Processes. The optimal setting of (n, h, k) of can be obtained through the economic design of the control charts to minimize the expected cost in unit time. Finally, a real case of STP of Software Life Cycle provided by a software company in Taiwan is utilized to demonstrate the effectiveness of the proposed procedure.