透過您的圖書館登入
IP:3.15.27.232
  • 學位論文

視覺化決策分析系統之建構

Development of a Visualized Decision Analysis System

指導教授 : 蔡榮發

摘要


資料包絡分析法(Data Envelopment Analysis,DEA) 與層級程序分析法(Analytic Hierarchy Process,AHP),是決策評估常用的分析方法。兩者適用之情況不同,若已有客觀的數據資料可供使用,則選擇DEA方法較適合。若沒有客觀的數據資料可供使用,僅能靠決策者主觀意見做決定時,則選擇AHP方法較適合。但是,兩方法在評估的過程中,會遇到一些問題,常造成決策者相當多的困擾。例如:(1)在DEA中,針對每個決策單元(Decision Making Unit,DMU)所評估出的效率值,有時會計算出多個有效率DMU(效率值為1),而造成無法對這些DMU進行有效之排名,產生鑑別度(Discriminate)的問題。(2)在AHP中,所產生的成對比較矩陣,有時會有偏好不一致性的問題,這時,必須修正成對比較矩陣,使不一致性程度能夠降低。故針對上述問題,本研究提出解決方法,並配合開發軟體工具以供決策者使用。其方法為:(1)利用Obata & Ishii(2003)所提模式,修改成本研究所使用的「正規化權重模式(Weight Normalization Model)」,並針對有效率DMU進行鑑別,其鑑別後結果可和DEA結果相結合。(2)利用Genest & Zhang(1996)所提「高爾圖(Gower Plot)」圖形,將成對比較矩陣中的方案偏好情形,繪製成數值型(Cardinal)與次序型(Ordinal)高爾圖,以分別比較方案間偏好,在程度上與次序上是否趨於一致。最後,本研究會舉例子來說明所提出方法與建構系統之可行性與正確性。

並列摘要


Data Envelopment Analysis(DEA) and Analytic Hierarchy Process(AHP) are useful decision analysis tools that can help decision makers to make a decision. DEA is suitable to evaluate objective data and AHP assists decision makers to assess the subjective data. Although these two methods are quite promising in measurement and ranking, some problems still exist. For instance, DEA has the problem of discrimination and may derive too many efficient DMUs. Accordingly, decision makers can not totally rank all DMUs. The pairwise comparison matrix of AHP also has the problem of inconsistency. To resolve the problem, the pairwise comparison matrix has to be adjusted iteratively until the consistency ratio fulfils the requirement. Therefore, this study develops a visualized decision analysis system to assist in dealing with these problems. First, weight normalization model is presented to treat the discrimination problem of DEA. Then, "Gower Plot" proposed by Genest and Zhang(1996) is adopted to show the preference of the alternatives graphically. In addition, we can easily check the consistency of the alternatives by the cardinal and ordinal Gower plots. Numerical examples are also presented to demonstrate the feasibility and applicability of the proposed methods and the developed system.

參考文獻


[1]高強,黃旭男,Toshiyuki Sueyoshi,管理績效評估資料包絡分析法,台北:華泰文化事業股份有限公司,2003。
[2]孫遜,資料包絡分析法:理論與應用,台北:揚智文化事業股份有限公司,2004。
[13]黎漢林,馬麗菁,黃芸珊,「決策方案的排序與分群─高爾圖及決策球的應用」,中國工業工程學刊,第二十一卷,第二期,2004,第113-120頁。
[5]普悠瑪數位科技,Visual C++ 6範例教本,台北:學貫行銷股份有限公司,2002。
[12]T. L. Saaty, The Analytic Hierarchy Process-Planning, Priority Setting, Resource Allocation, New York: McGraw-Hill, 1980.

延伸閱讀