DEAHP cross-efficiency model based on rank reversal: An application to layout planning of Taoyuan International Airport Terminal 3

• Main Authors: Chuang, Jiun-Yau, 莊鈞堯
• Other Authors: Ming-Miin Yu
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/ne5y2f

碩士 === 國立臺灣海洋大學 === 運輸科學系 === 105 === In the past, people usually used a single criterion to make a decision, but they can only solve the relatively simple problem. However, it is full of composite problem in present, so it must give different weights in different criteria when Multi Criteria Decision Method (MCDM) is used to assess decision-making schemes. In many decision-making methods, there are many ways to rank the pros and cons of the alternatives. In the process, the concept of hierarchy is used to calculate weight of each alternative. AHP(Analytic Hierarchy Process)has been one of method that be used widely. There are three purpose of AHP: systematizing complex problems, resolving criteria for different hierarchies and decreasing decision error. The evaluation method compares alternatives or criteria, gets score to calculate weight, and then ranks alternatives. In MCDM, Ramanathan (2006)have used DEA to evaluate weight of AHP, called DEAHP, but there are some drawbacks. For example, the multiplier could be zero, and the result confined to self- evaluation. Afterwards, Wang et al. (2008) used the concept of assurance region (AR) to resolve multiplier problem, but self-evaluation wasn’t resolved. Lee (2016) used DEAHP cross-efficiency method to solve these two main drawbacks. In addition, rank reversal is always a controversial problem when AHP is used to evaluate alternatives. Hence, in order to obtain the discriminative and persuasive results, this research considers this problem. In empirical analysis, this research evaluates the layout planning of Taoyuan International Airport Terminal 3 under construction. This research uses the importance of facility placement to rank the three shortlisted alternatives. Next, under the consideration of the problem of rank reversal, a dummy alternative (alternative 4) which is not announced is added to explore whether the weight of alternative is varied or not.