ABC inventory classification in the presence of both quantitative and qualitative criteria S.A. Torabi ⇑ , S.M. Hatefi, B. Saleck Pay Department of Industrial Engineering, College of Engineering, University of Tehran, Tehran, Iran article info Article history: Received 28 June 2011 Received in revised form 13 March 2012 Accepted 19 April 2012 Available online 3 May 2012 Keywords: ABC inventory classification Data envelopment analysis Common weights Qualitative criteria abstract Organizations classically employ the ABC analysis to have an efficient control on a large number of inven- tory items. The customary classification method considers just one criterion, i.e., the annual dollar usage to classify inventory items. Recently, several methods have been developed for ABC inventory classifica- tion, especially DEA-like models that account for other important criteria leading to more logical results in practice. However, these models assume that all criteria are of quantitative type and hence cannot han- dle the qualitative criteria which are not stated numerically but as linguistic terms. To alleviate this shortcoming, this paper proposes a modified version of an existent common weight DEA-like model by using of some concepts in the current imprecise DEA (IDEA) models and then applies it for ABC inventory classification in the case where there exist both quantitative and qualitative criteria. The merits of employing the modified model to solve the multi criteria inventory classification (MCIC) problem are dis- cussed. A case example is also illustrated to demonstrate the applicability of the modified model in the context of MCIC problem as well as its superiority over existing approaches. Ó 2012 Elsevier Ltd. All rights reserved. 1. Introduction ABC analysis is an efficient and most widely employed tech- nique for inventory classification in organizations. This technique that is based on Pareto principal is an easy to understand method that divides the inventory items into the three classes according to specific criteria. Class A, commonly are those items of high impor- tance but few in numbers, class C, less important but large in num- ber and class B is between this two classes. More discussion about the suitable inventory control policies for each class of items can be found in Silver, Pyke, and Peterson (1998). Conventional ABC analysis accounts for only one criterion, mostly ‘‘annual dollar usage’’, for classification of inventory items. However, there are too many other criteria (both quantitative and qualitative) that may significantly affect this classification such as: inventory holding unit cost, part criticality, length and variability of replenishment lead time, commonality, substitutability, scarcity, durability and stock-out unit penalty. Therefore, traditional classi- fication method cannot provide trusty results (Guvenir & Erel, 1998) and including other criteria (especially those of qualitative ones) in the decision process is of particular interest. To solve such a MCIC problem, several models have been proposed. Bhattach- arya, Sarkar, and Mukherjee (2007) and Rezaei and Dowlatshahi (2010) provide comprehensive reviews on the various methods realized for MCIC issue in the literature. The rest of the paper is organized as follows. The related litera- ture review is investigated in Section 2. In Section 3, problem def- inition and a brief discussion about qualitative data are first presented, and then the modified model is elaborated. Application of the strong ordinal relations and their role in the modified model are presented in Section 4. In Section 5, applicability of the modi- fied model is demonstrated through a numerical example taken from the literature. The comparative results with those of provided by previous approaches are also illustrated in this section. Finally, concluding remarks are reported in Section 6. 2. Literature review We review the relevant literature into the three different streams: the models incorporating just quantitative criteria, the models taking into account both quantitative and qualitative crite- ria, and some relevant DEA-based models that can be applied for MCIC problem. 2.1. Models incorporating only quantitative criteria In this sub-section, we review those models that only account for quantitative criteria when solving the MCIC problem. In this stream, there are several meta-heuristic based works (such as ge- netic algorithms and particle swarm optimization) and other arti- 0360-8352/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.cie.2012.04.011 ⇑ Corresponding author. E-mail addresses: satorabi@ut.ac.ir (S.A. Torabi), smhatefi@ut.ac.ir (S.M. Hatefi), bsaleckpay@gmail.com (B. Saleck Pay). Computers & Industrial Engineering 63 (2012) 530–537 Contents lists available at SciVerse ScienceDirect Computers & Industrial Engineering journal homepage: www.elsevier.com/locate/caie