Abstract

MENDEZ GIRALDO, Germán Andrés  and  LOPEZ SANT, Edüyn Ramiro. Methodology to demand forecastingunder multiproduct environments and high variability. Tecnura [online]. 2014, vol.18, n.40, pp.89-102. ISSN 0123-921X.

Abstract This paper provides a methodology for the demand management in multiproduct environments for several stages in a supply chain with high variability in the demand. The methodology proposed consists of two phases. The first phase looks for products classification according to two criterions, the first is the movement factor representing inventory's turns and the second is the importance factor given by variables such as cost, volume and weight. The output of this phase is a product category to identify the most important products and which should be discontinued. The second phase consists in performing an aggregate forecast according to the variability in each category. We use the coefficient of variation (CV) as measure of variability, and the double exponential smoothing method and the double moving average method as forecasting techniques, although the methodology is flexible to use another time series methods. The methodology was implemented in a leader company in Colombia of poly vinyl products achieving an improve of 10% in the mean absolute deviation, approximately. On the other hand the average inventory in the system was reducing significantly. Finally this work is a first step for a dynamic classification of goods in order to manage inventory holistically a multistage supply chain.

Keywords : ABC Classification; Demand Forecasting; Production Planning; Time Varying Systems.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )