TY - CHAP
T1 - Efficiency measurement in data envelopment analysis with fuzzy data
AU - Kao, Chiang
PY - 2015/1/1
Y1 - 2015/1/1
N2 - Conventional data envelopment analysis (DEA) requires the data to have crisp values, which can be measured precisely. However, there are cases where data is missing and has to be estimated, or the situation has not occurred yet and the data has to be predicted. There are also cases where the factors are qualitative, and thus the data cannot be measured precisely. In these cases, fuzzy numbers can be used to represent the imprecise values, and this paper discusses the corresponding measurement of efficiency. Based on the extension principle, two approaches are proposed; one views the membership function of the fuzzy data vertically, and the results are represented by membership grades. The other views it horizontally, and the results are represented by α-cuts. The former approach is easier to understand, yet is applicable only to very simple problems. The latter, in contrast, can be applied to all problems, and is easier to implement. An example explains the development and implementation of these two approaches.
AB - Conventional data envelopment analysis (DEA) requires the data to have crisp values, which can be measured precisely. However, there are cases where data is missing and has to be estimated, or the situation has not occurred yet and the data has to be predicted. There are also cases where the factors are qualitative, and thus the data cannot be measured precisely. In these cases, fuzzy numbers can be used to represent the imprecise values, and this paper discusses the corresponding measurement of efficiency. Based on the extension principle, two approaches are proposed; one views the membership function of the fuzzy data vertically, and the results are represented by membership grades. The other views it horizontally, and the results are represented by α-cuts. The former approach is easier to understand, yet is applicable only to very simple problems. The latter, in contrast, can be applied to all problems, and is easier to implement. An example explains the development and implementation of these two approaches.
UR - http://www.scopus.com/inward/record.url?scp=84955116098&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84955116098&partnerID=8YFLogxK
U2 - 10.1007/978-1-4899-7553-9_12
DO - 10.1007/978-1-4899-7553-9_12
M3 - Chapter
AN - SCOPUS:84955116098
T3 - International Series in Operations Research and Management Science
SP - 341
EP - 354
BT - International Series in Operations Research and Management Science
PB - Springer New York LLC
ER -