Weights based clustering in Data Envelopment Analysis using Kohonen Neural Network: An Application in Brazilian Electrical Sector

This paper presents a methodology developed using the techniques Data Envelopment Analysis (DEA) and Self-Organizing Maps (SOM) in order to cluster productive units under analysis. In this study, the input vectors are the inputs and outputs contributions from DEA in order to generate groups with sim...

Full description

Saved in:
Bibliographic Details
Published in:Revista IEEE América Latina 2015-01, Vol.13 (1), p.188-194
Main Authors: Araujo Alves, Laura, Carlos Correia Baptista Soares Mello, Joao
Format: Article
Language:eng ; por
Subjects:
Biological system modeling
Computational modeling
Data envelopment analysis
DEA
Linear programming
Mathematical model
Organizing
Self-Organizing Maps
Software
Vectors
Weights Contribution
Citations: Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This paper presents a methodology developed using the techniques Data Envelopment Analysis (DEA) and Self-Organizing Maps (SOM) in order to cluster productive units under analysis. In this study, the input vectors are the inputs and outputs contributions from DEA in order to generate groups with similar profiles considering the relevance of selected variables. This way, this clustering is different from most part of the applications found in literature, which commonly use the efficiency scores assessed by DEA as input vector. For this purpose, two processes are incorporated into the methodology to apply the method: the weights used are converted into the contribution of each variable to the DMU and, in addition, a problem of linear programming is used to determine which set of weights from the optimal weights generated by DEA will be used as input vector of SOM.
ISSN:1548-0992
1548-0992
DOI:10.1109/TLA.2015.7040647