A data envelopment analysis (DEA) evaluation method based on sample decision making units

Quanling Wei, Hong Yan

Research output: Journal article publicationJournal articleAcademic researchpeer-review

21 Citations (Scopus)


Most of evaluation methods on large number of candidates are based a single criterion. To bring the multiple attribute evaluation method Data Envelopment Analysis (DEA) into evaluating large number of elements, it needs to set up the performance standards and an evaluation procedure by the DEA model. In this paper, we first determine a set of "standard" candidates, called in decision making units (DMUs) in the DEA terminology. This standard set is called "training set". We then establish the evaluation procedure based on this "training set" for measuring a large number of DMUs. We first investigate the efficiency evaluation of a new DMU along with the original definition based on the sum formed production possibility set which is formed by the n DMUs in the training set and the new DMU. We then identify the intersection form of the production possibility set formed only by the n DMUs from the training set. And show that the new DMU evaluation is simply to check if the DMU satisfies a set of linear inequalities. The intersection formed production possibility set formed by the n DMUs from the training set is fixed for evaluating any new DMU. Therefore, it provides an efficient and effective method for dealing with a large amount of data. The method can be regarded as a complementary approach for data mining.
Original languageEnglish
Pages (from-to)601-624
Number of pages24
JournalInternational Journal of Information Technology and Decision Making
Issue number4
Publication statusPublished - 1 Jul 2010


  • Data Envelopment Analysis
  • data mining
  • intersection form
  • production possibility set

ASJC Scopus subject areas

  • Computer Science (miscellaneous)


Dive into the research topics of 'A data envelopment analysis (DEA) evaluation method based on sample decision making units'. Together they form a unique fingerprint.

Cite this