CLUSTERING APPROACHES FOR MANAGING SIMILAR ITEMS IN THE FIELD OF IRREGULAR AND SPORADIC DEMAND PROFILES: EVIDENCE FROM AN EMPIRICAL COMPARISON

Managing irregular and sporadic demand patterns is a fundamental task in several real life contexts, such as spare parts consumption, multi-echelon supply chains or start-up production. This work is a study of re-order policies and stock inventory management approaches aimed at optimizing pre-defined performance indexes. Specifically, given a firm operating in the field of electric resistance manufacturing, the focus is on the application of different item clustering methods, in order to define groups of items with similar behavior that require similar management approaches. The work offers a framework for the comparative evaluation of two different item clustering methods, by means of a simulative approach, available when product demand profiles are irregular and sporadic. In order to compare them, a multi-criteria technique is preferable because of the high uncertainty of the cost structure. Hence, after running the simulation, three Key Performance Indicators (KPI) are estimated: the average inventory level, the average number of backorders that occurred and the average number of emitted orders. Finally, some conclusions are drawn by defining a field of implementation for each clustering approach studied.