These settings specify the behaviour of the model when trying to find a relationship between load and weather as well as the estimation of load changes that have occurred between the reference periods and the base period. The models use stepwise multiple regression with a backward glance for the model selection. From temperature, wind and global radiation as primary predictors, by default, 11 secondary predictors are constructed and used as input data into the regression. The user can override these settings and explicitly specify which primary input data that should be used and how to construct the secondary predictors. The final weather predictors are expressed as the difference between the values during the base period and the corresponding day type pattern values. These differences are sent to the regression together with the estimated day type pattern for the load.