Metodología para pronosticar demanda y clasificar inventarios en empresas comercializadoras de productos mayoristas

Objective: To recommend a methodology that allows for inventory classification and demand forecasting, by wholesale supplier companies, as critical factors to implement performance optimization. Methods and techniques: The methodology relies on the use of a multilayer artificial neural network developed with Weka software, which adds a solution to inventory item classification problems, which is based on ABC and analytics of hierarchy processes (AHP). The methodology was developed in three phases, the first one was in charge of inventory classification, the second was related to forecasting, and the third, to integrated result analysis. Main results: A hierarchical scale of variables was suggested for inventory item classification, as well as weigthing opinions and sub-opinions, and its selection extent. An effective way of forecasting individual demands was presented for every inventory item. Conclusions: The application of this methodological tool by ACINOX sales company in Holguin province corroborated its effectiveness to solve inventory classification problems and demand forecasting. As a result, all the executives have access to a tool that contributes to decision-making, in order to favor better items classification and forecasting. Key words: demand forecasting; aggregate planning; artificial neural networks; inventory classification; ABC classification.