DEASort: assigning items with data envelopment analysis in ABC classes

Multi-criteria inventory classification groups similar items in order to facilitate their management. Data envelopment analysis (DEA) and its many variants have been used extensively for this purpose. However, DEA provides only a ranking and classes are often constructed arbitrarily with percentages. This paper introduces DEASort, a variant of DEA aimed at sorting problems. In order to avoid unrealistic classification, the expertise of decision-makers is incorporated, providing typical examples of items for each class and giving the weights of the criteria with the Analytic Hierarchy Process (AHP). This information bounds the possible weights and is added as a constraint in the model. DEASort is illustrated using a real case study of a company managing warehouses that stock spare parts.