Abstract
ABC classifications can be constructed based on a wide range of approaches (varying from formal multi-criteria optimization models to more subjective approaches like the analytic hierarchy processes). Several multi-criteria inventory classification (MCIC) models in particular have recently been proposed in the academic literature. However, even in the presence of the same criteria and the same input data, these models may offer considerably different ABC classifications. That is, for example, an item that is classified based on a certain model in the most important class (class A for example) could be classified by another model in the least important class (class C for example). Such discrepancies have motivated the work described in this paper the objective of which is to tackle the inconsistent operation of existing models for ABC classifications. Based on two well-known MCIC models proposed in the literature, we first develop a new hybrid model which succeeds in reducing the conflict in the resulting classifications. We then propose some new procedures that may take as an input differing ABC classifications to reach as an output a consensus between them. The efficiency of such procedures is evaluated through a number of computational experiments performed on a dataset commonly used in the inventory literature. Using such a dataset allows us to contrast our results with those reported by other researchers.