路線別模糊資料包絡分析模式評估運輸服務效率

Abstract

本論文研提新型資料包絡分析(DEA)模式評估運輸服務效率，主要針對路線別績效分析及納入模糊變數時進行DEA模式建構。首先，針對路線別績效分析提出路線別資料包絡分析(RDEA)模式。RDEA模式包含兩類模式，分別為RCCR與RBCC模式。此兩類路線別模式均可同時分析公司與路線別之績效值。其中，RDEA模式是利用三階段求解法求算公司績效值、路線績效值以及最佳共同成本配置比例。本研究驗證三階段求解之績效排序一致性，而後利用台灣城際公路客運公司之績效評估驗證RDEA模式之適用性。 其次，針對模糊變數提出整合式模糊資料包絡分析(IFDEA)模式。IFDEA模式包含兩類型之模式，分別為IFCCR與IFBCC模式，分別用於衡量固定規模報酬及變動規模報酬下之效率值。IFDEA模式利用績效值上下限整合概念，可針對模糊變數之上下限進行不同的差額變數分析。本論文亦利用相同之公路客運案例驗證IFDEA模式之適用性。 最後，經整合RDEA模式與IFDEA模式進而提出兩類型之整合式路線別模糊資料包絡分析(IRFDEA)模式，分別為IRFCCR與IRFBCC模式。IRFDEA模式可同時考量路線別與模糊變數，此模式利用RDEA模式之概念提出，因此IRFDEA模式亦為三階段之整合模式，如同RDEA模式可同時求解公司績效值、路線績效值及最佳共同成本配置比例。同時，IRFDEA模式亦保有公司別與路線別績效一致性之特性，最後亦利用相同之公路客運案例驗證IRFDEA之適用性。

This study proposes three different types of data envelopment analysis (DEA) modeling to remedy two research gaps: lacking of route performance evaluation and including vagueness of some variables measurement. First, route-based data envelopment analysis (RDEA) modeling is proposed. This study develops two novel RDEA models, termed RCCR and RBCC, that jointly measure the route-level and company-level efficiencies amongst transport carriers. The core logics comprise a three-stage procedure that determines company efficiency, route efficiency and optimal allocation ratios for the common inputs. We prove that the ranking order of company performance determined by the route-based DEA model is identical to that determined by the company-based DEA model. An empirical study of intercity bus transport companies in Taiwan demonstrates the superiority of the proposed models in identifying the less efficient routs/companies as well as in reducing the input slacks without subjective conjectures. Second, integrated fuzzy data envelopment analysis (IFDEA) modeling is proposed. This study develops two IFDEA models, termed IFCCR and IFBCC, by combining both lower- and upper-bound efficiency frontiers into a single one under a specific α-cut. The proposed IFDEA models can simultaneously determine the slack values for both lower- and upper-bound input/output variables. A numerical example shows that the proposed IFDEA models are more generalized and have greater simplicity than an existent FDEA model. An empirical study of the same case further demonstrates the superiority of the proposed IFDEA models, which have successfully dealt with both quantitative (crisp) and qualitative (fuzzy) variables. Third, integrated route-based fuzzy data envelopment analysis (IRFDEA) modeling is proposed. This study develops two IRFDEA models, termed IRFCCR and IRFBCC, which jointly measure the route-level and company-level efficiencies with both crisp and fuzzy variables. The proposed models also comprise three stages. The first stage uses an integrated company-based IFDEA model to acquire a set of optimal multipliers. The second stage uses the solved multipliers to determine its optimal allocation ratios for the common inputs among the routes within a company to maximize the efficiency of all routes. The third stage further determines the relative efficiency for all routes across the companies. An empirical study of the same case demonstrates the superiority of the proposed models in pining down the less efficient routes/companies and in suggesting how much the inputs of less efficient routes/companies should be improved.