Experimental comparison of results provided by ranking methods in Data Envelopment Analysis

Autor: Miłosz Kadziński, Anna Labijak-Kowalska
Rok vydání: 2021
Předmět:
Zdroj: Expert Systems with Applications. 173:114739
ISSN: 0957-4174
DOI: 10.1016/j.eswa.2021.114739
Popis: We consider the problem of ranking Decision Making Units (DMUs) in Data Envelopment Analysis. We illustrate the use of fifteen selected approaches on a numerical example. They represent different categories, including cross- and super-efficiency, multivariate statistics, decision analysis, benchmarking, virtual DMU, and social networks. Moreover, we formalize a new category of ranking methods based on the concept of Robustness Analysis. They exploit a space of feasible input/output weight vectors with the Monte Carlo simulation to derive the expected efficiencies or ranks, or to compute the priorities or net flow scores of DMUs based on the matrix of pairwise efficiency outranking indices. The rankings constructed by all methods are compared on both artificially generated and real-world datasets with different numbers units, inputs and outputs, and performance distributions. The considered datasets represent the most common application areas of the DEA methods, such as finances, education, transportation, healthcare, farming, and the energy industry. The results are quantified in terms of five measures. We indicate that the choice of a method has a significant impact on the ranking, revealing the procedures that offer similar results or differ vastly in terms of the recommended order or the most preferred DMU.
Databáze: OpenAIRE
Externí odkaz: