A model for high-frequency load data via double Holt-Winters with exogenous explanatory variables

Abstract

This work presents an extension of the Holt-Winters model with multiple cycles by incorporating exogenous explanatory variables. We apply the model to hourly load data in Southeast Brazil. Temperature data collected at weather stations in the city of Rio de Janeiro are used as explanatory variables, both in the original and transformed scales, as HDDs and CDDs. The inclusion of the temperature variables improves the predictive power of the model, generating predictions with greater accuracy.