建構軟體發展流程之多變量管制系統

Construct the Multivariate Control System for Software Development Process

陳啟揚(Chi-Yang Chen)

指導教授 ： 唐麗英

國立交通大學/管理學院/工業工程與管理系所/碩士(2008年)

https://doi.org/10.6842/NCTU.2008.00018

摘要

近年來能力成熟度整合模式(Capability Maturity Model Integration, CMMI-SE/SW/IPPD/SS)已經成為軟體產業品質保證及國際合作的認證標準，其中CMMI Level 4所要求的數量化管理必須使用統計製程管制(statistical process control, SPC)來監控軟體發展流程(software development process, SDP)，為了使得軟體發展流程具有穩定性與可預測性，越來越多軟體廠商使用管制圖來監控軟體發展流程。雖然製造業已經成功的運用管制圖來偵測產品或製程變異，但是軟體發展流程與製造流程彼此之間卻有相當大的差異，使得軟體廠商不易成功地將SPC應用於軟體發展流程上。由於軟體指標不僅一個，且彼此有關連性，因此本論文之主要目的是發展一套多變量管制系統來同時監控多個軟體指標。本論文最後利用台灣某軟體公司的實際資料來說明本論文之管制系統。

關鍵字

多變量管制圖 ； 軟體發展流程 ； 軟體指標

並列摘要

In the recently year, the recognizing of Capability Maturity Model Integration (CMMI) have became insurance of quality of a software industry and standard of international cooperation. The level 4 of CMMI called Quantitatively Managed Level requires using statistical process control(SPC) to monitor software development process(SDP). In order to make the SDP more stable and more predictable, more and more software companies start using control chart to monitor SDP. Although manufacturing industry have use control chart to detect variation of products or variation of process successfully for years, the difference between SDP and manufacturing process make it is hard to use control chart on SDP successfully for software companies. Since there are more than one software metrics need to be monitored and one metric correlates with another metric, this study propose a multivariable control system to monitor more than one metrics at the same time. A case study is also presented to verify this system does work.

並列關鍵字

CMMI ； multivariable control chart ； software development process ； software metrics

參考文獻

吳炤華，2000，「積體電路生產線上結合缺陷數與缺陷群聚之Hotelling T