有效的評估過程在管理上對整個專案的進行成效具有關鍵性的影響,因此,績效評估是軟體專案管理中一項重要的研究議題。至目前為止,軟體專案績效評估尚缺乏一套完整與系統化的方法,往往造成專案管理者無法確實瞭解專案的情況,導致專案面臨中止或無法符合高階主管預期的情形發生。有鑑於此,本研究根據績效評估理論建立一套以流程為導向的軟體專案績效評估模式,其評估指標包括時程、成本、品質、生產力與智慧資本,並結合專案目標達成度評估法、圖表評等尺度法、生產力比率評量法及層級分析法等績效評估方法做為評估指標制訂之基礎,據以發展一個軟體專案績效評估系統,同時,結合智慧型代理人協助專案之狀態評量及提供相關決策方案之建議,供專案管理者進行決策之用途。 為驗證本研究所發展之軟體專案績效評估系統在實務上應用之可行性與實用性,以國內某公司之資訊部門作為實證對象。本研究實證結果如下:(1)設計一軟體專案績效評估指標應考慮因素中,排行榜由高至低為:須符合軟體專案的特性且反應真實的情況、完整性、與專案控管目標的一致性、定期的衡量以及追蹤。(2)優良的績效評估指標特性項目中,排行榜由高至低為:現實性、客觀性、具體性、可衡量、實現性、有時限性、簡單性。(3)本研究所發展的績效評估指標所具備的特性中,排行榜依序為:客觀性、有效性、可行性、量化、完整性、實用性。(4)推動軟體專案績效評估工作的關鍵成功要素前三名分別為:高階主管的支持、專案相關人員認同績效評工作並願意高度配合、評估指標是經過上屬與下屬持續的溝通過程。(5)一個軟體專案績效評估系統在評估過程中所帶來的效益,前二名分別為:協助管理者進行績效評估工作,提升評估作業的效率與品質、有效掌控專案活動的進行、落實績效評估制度於實際的作業當中及使專案人員與專案管理者雙方均有一致的目標。然而,本研究結果可以提供實務界在進行軟體專案績效評工作之參考用途。
Performance evaluation is considered as has significant inference in effectively managing a software project. Therefore, performance evaluation becomes an important research topic in software project management. So far, few of research papers propose a comprehensive and systematic method of performance evaluation. Many software projects are terminated or unexpected results are occurred due to lacks of information about the status of a project. To resolve the issues mentioned above, this research proposes a process-oriented performance evaluation model for software projects. Five kinds of performance index including schedule, cost, quality, productivity and intellectual capital are constructed. In this study we use computational methods such as management by objectives (MBO) approach, rating scale method, productivity ratio method, and analytic hierarchy process (AHP) during the performance evaluation process. Based on the research model, we develop an agent based software project performance evaluation system to support the assessments of project status and decision making for a project manager. To examine the feasibility and applicability of research model, an empirical study is conducted in a high tech company in Taiwan. Results of this research are shown as follows: (1) ranking of importance of performance index are reflection of project status, completeness of performance evaluation, accordance with the objectives of software projects, and appliances of measurement and follow up in regular time; (2) ranking of importance of good performance index are applicability, objectiveness, measurability, and simplicity; (3) ranking of attainability of performance index proposed in this research are objectiveness, effectiveness, feasibility, quantification, completeness, and applicability; (4) ranking of importance of critical successful factors of performance evaluation are commitments by management, involvements and supports from project teams, and communication between project teams during the process of developing the performance index; (5) ranking of benefits of implementations of performance evaluation are improvements of efficiency and quality of performance evaluation, and effectively attain the objectives between the project leader and team members. In summary, findings of this research provide useful guidelines for academic research and practices.