We thank the discussants for their comments and for their insights into the problems of electricity load modeling and forecasting. The discussants provided many proposals for modifying or extending our load model that are worthy of investigation. We emphasize that our work is ultimately intended for use in industrial applications and reflects the requirements of these applications: producing accurate forecasts quickly, with robustness to irregularities in the data and adaptability to changing conditions with minimal need for calibration data. These requirements influence both the structure of our model and our reaction to the discussants' proposals. For example, we do not expect to have the luxury of fitting the model to several years of historical data, as suggested by Espada and Durban, and even the use of the forecasting method of Martinez-Alvarez et al., as suggested by Aung, is not feasible for the OlyPen data because the method is based on the analysis of load curve patterns in historical data and effectively requires a full year of historical data for calibration.
展开▼
