Selección y utilización de niveles de desagregación adecuados en pronósticos de series temporales: Caso de estudio en una empresa de suscripción utilizando el proceso analítico jerárquico

Abstract

Hierarchical aggregation/disaggregation of time series in order to make forecasts is a frequent challenge in business and econometric scenarios. This work presents a novel approach for selecting an adequate time series disaggregation level as a starting point for making forecasts. The methodology combines qualitative criteria -such as business resourcesand decision environment- and quantitative criteria -such as information quality and forecast ability- in a multicriteria decision making task which is addressed through the analytic hierarchy process (AHP) technique. Results from a study case in a subscription business model company show the usefulness of combining AHP and time series forecasting techniquesand the importance of multicriteria decision-making in the task of selecting an adequate aggregation/ disaggregation level.