Table 1 Required inputs from the user for baseline model fitting

Length of the training period

Number of years, number of observations

Retrospective : All data

Prospective : Half dataset

Purge of the training period

Data above selected percentile, above cut-off value, or in user-defined periods

Above the 15% highest percentile

Regression equation

Linear, quadratic or cubic terms. Annual, semi-annual or quarterly periodicity

Automated model selection

Upper forecast limit (UFL)

Percentile between 50% and 100%

95%

Minimum duration above UFL defining an unexpected change

Number of observations

14 days/2 weeks/1 month