ROBUST METHODS APPLIED TO ENERGY‐CONSUMPTION DATA

In many instances a model is based on nonexperimental data and is used for prediction. In this situation, using both the least-squares and the robust methods simultaneously can result in models with improved predictive performance.