A non-compensatory classification approach for multi-criteria ABC analysis

Abstract

ABC analysis is a widespread inventory management technique designed to classify inventory items—based on their weighted scores—into three ordered categories A, B and C, where category A contains the most important items and category C includes the least important ones. This paper proposes a new ABC classification approach which involves a non-compensatory aggregation procedure, based on a simplified ELECTRE III method, to compute the score of each inventory item. A non-compensatory aggregation scheme means that the bad scores of an item on some significant criteria could not be offset by its high performances on the other criteria. This way of proceeding prohibits this kind of items from being classified into good categories and therefore generates a more realistic ABC classification of inventory items. Since the application of the simplified ELECTRE III method requires the knowledge of some parameter values, the continuous variable neighborhood search meta-heuristic will be used for their estimation. The comparative study—conducted on two real datasets—shows that the classification of items produced by our proposed approach has generated the lowest inventory cost value among those produced by all tested classification models.