A subset of consumers will in general have a different hourly day type pattern then than all consumers have. If, for example the forecast series contains a disproportionate number of industrial, the day type patterns will differ from the auxiliary series in one way during working days and in another way during the weekends. In the calendar settings, the user can specify which day type calendar should be used when determining the relations between the day type patterns, or if a simpler distinction between days is sufficient.
.
In ”Days/year”, the number of days used each year should be specified.
The period used is centered around the date for
the forecast start date for years back and placed immediately before the last
received outcome for the forecast series the current year Data will be used the
number of years back specified in the other box. The same year, only half of the
days specified under “Days/year” are used. All days during this reference period
are grouped by day type and for each day type average hourly patterns are
constructed, both for the forecasted and auxiliary series and then the hourly
quotient between these two series are calculated. If not any weather dependent
adjustments are estimated, these day type quotients are applied on the
forecasted values for the auxiliary series in order to obtain the final forecast
values. Also, daily hourly averages for a number of weather parameters are saved
for each day type found, to be used in the next step in the model estimation
process.