Electric load forecasting is a complex problem that is linked with social activity considerations and variations in weather and climate. Furthermore, electricity is one of only a few goods that has almost no inventory. The absence of inventory means that electric utilities’ operations depend heavily on accurate load forecasts to balance the demand and supply in an almost real-time fashion. This electric load forecasting problem is even more challenging for holidays, which have limited historical data and varying demand patterns. These challenges result in the forecast error for holidays being higher, on average, than it is for regular days. This paper investigates three practical holiday electric demand forecasting techniques: modeling holidays as weekends, modeling holidays using holiday dummy variables, and a two-stage method in which the second stage models residuals. The empirical results from this investigation show that the selection of holiday electric demand forecasting strategy depends both on the holiday itself and on the availability of the historical data.
SAS employees Jingrui Xie and Alex Chien explain electric load forecasting in their SAS Global Forum 2016 paper.