Loading…

Data envelopment analysis of randomized ranks

Probabilities and odds, derived from vectors of ranks, are here compared as measures of efficiency of decision-making units (DMUs). These measures are computed with the goal of providing preliminary information before starting a Data Envelopment Analysis (DEA) or the application of any other evaluat...

Full description

Saved in:
Bibliographic Details
Published in:Pesquisa Operacional 2002-12, Vol.22 (2), p.203-215
Main Author: Sant'Anna, Annibal P.
Format: Article
Language:English
Subjects:
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Probabilities and odds, derived from vectors of ranks, are here compared as measures of efficiency of decision-making units (DMUs). These measures are computed with the goal of providing preliminary information before starting a Data Envelopment Analysis (DEA) or the application of any other evaluation or composition of preferences methodology. Preferences, quality and productivity evaluations are usually measured with errors or subject to influence of other random disturbances. Reducing evaluations to ranks and treating the ranks as estimates of location parameters of random variables, we are able to compute the probability of each DMU being classified as the best according to the consumption of each input and the production of each output. Employing the probabilities of being the best as efficiency measures, we stretch distances between the most efficient units. We combine these partial probabilities in a global efficiency score determined in terms of proximity to the efficiency frontier. Probabilidades e chances relativas são aqui comparadas como medidas de eficiência de unidades tomadoras de decisão (DMUs). Avaliações de preferência, qualidade e produtividade costumam ser medidas com erros e estar sujeitas à influência de outras perturbações aleatórias. Reduzir as avaliações iniciais a postos e tratar estes como estimativas de parâmetros de locação de variáveis aleatórias permite calcular as probabilidades e chances relativas de cada opção ser classificada como a de maior preferência. Esta transformação amplia as distâncias entre as DMUs mais eficientes. As probabilidades e as razões de chances relativas delas derivadas podem ser combinadas em termos de proximidade à fronteira de excelência. Aqui se apresenta evidência de que os escores de eficiência derivados das probabilidades e chances relativas são mais correlacionados com as medidas que combinam que os escores derivados dos postos ou das razões de produtividade.
ISSN:0101-7438
1678-5142
0101-7438
DOI:10.1590/S0101-74382002000200007