A Two-Level Induced OWA Procedure for Ranking DMUs Under a DEA Cross-Efficiency Framework

Amar Oukil*

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingChapter

2 Citations (Scopus)

Abstract

Cross-efficiency (CE) evaluation is an extension of data envelopment analysis (DEA) used to fully rank decision-making units (DMUs). The ranking process is carried out on the matrix of CE scores, where an ultimate efficiency score is computed for each DMU through an adequate amalgamation process. In this paper, we propose a ranking procedure that computes the ultimate scores over two aggregation levels, which involve ordered weighted averaging (OWA) induced by different preference settings. First, the preference voting system embedded under the CE matrix is employed to derive aggregate votes as inner quantifiers of the consensual importance of the DMUs. Next, the preference order induced from the aggregate votes is adopted as a ground for rearranging the rows of the original CE matrix prior to the ultimate calculation of the efficiency scores. To substantiate the impact of subjectivity on the structure of the ranking patterns and, hence, assess the robustness of the proposed procedure, the induced OWA is implemented with different optimism level values. The proposed methodology is applied for selecting the best material handling equipments (MHE) within a sample of 25 real-life MHEs, collected from catalogues of MHE manufacturers and vendors.

Original languageEnglish
Title of host publicationLecture Notes in Production Engineering
PublisherSpringer Nature
Pages495-521
Number of pages27
DOIs
Publication statusPublished - Jan 1 2023

Publication series

NameLecture Notes in Production Engineering
VolumePart F1164

Keywords

  • Aggregation
  • Data envelopment analysis
  • Induced ordered weighted averaging
  • Preference voting

ASJC Scopus subject areas

  • Industrial and Manufacturing Engineering
  • Economics, Econometrics and Finance (miscellaneous)
  • Safety, Risk, Reliability and Quality

Cite this