Filtering Time Intervals
The Filter time spans method primarily allows time series to be broken down into relevant and less relevant time ranges. The method includes two steps: first filtering, then replacing the Filter values.
As filter you define time intervals by selecting Years, Months, Weekdays, or Sundays/Holidays, Start and End times.
To determine a base load, the Weekdays of the time series are filtered out, and only Saturdays and Sundays are considered.
In another case, the time series of a heat demand is filtered so that only all Mondays, Tuesdays, and Thursdays from 08:00 to 20:00 and all Sundays and holidays from 18:00 to 00:00 the following day for the months January to April and September to December form a new time series.
The filter can be used in both the positive and negative sense: The user selects as Filter range whether the values which correspond to the filter criteria or the values which do not correspond to the filter criteria are replaced or deleted.
Optionally, the data points defined in the filter area are
Start Method
To start the method, select Methods → Filter → Filter time spans from the ribbon (see figure below) or in the bottom toolbaror the context menu of the time series in the Project Explorer.
In the window that opens (see following slider), select the time series to be edited with Select time series. A multiple selection is possible.
Public Holidays
To filter regional holidays, select the Country, Region, and, if necessary, the City. While public holiday calendars are stored in ETA for Germany, United Kingdom, and Australia, the selection Other serves to filter time series of the other countries. Only a time zone can be selected as the Region for the Country selection Other because the public holidays are not known and can therefore not be filtered out.
Filter Criteria
Under the four tabs Years, Months, Days, Time of Day, the filter criteria must be defined. The tabs are not optional, but the filter consists of the information under all four tabs. Therefore, all four form pages should be considered and all settings should always be checked. The following applies in particular:
The selection by public Holidays dominates the selection by Weekdays, that means: the selection Sunday/Holiday filters out all Sundays and holidays, the selection Tuesday only filters out all Tuesdays that are not holidays.
Under Filter range, you specify whether the filter criteria are to be applied as positive conditions or as exclusion criteria. Should the Filter values be replaced for the times that
Replacement Value
The filtered out time points can be assigned a replacement value or empty data-points (NaNs) or they can be deleted. If data points are deleted, an equidistant time series usually becomes a non-equidistant time series.
Finally, you select whether the respective existing time series should be overwritten with the filtered values or whether a new time series should be created in addition.
The weekdays are to be filtered from an equidistant time series.
Under the Days tab, place a check mark in front of Weekdays. This means that Monday, Tuesday, Wednesday, Thursday, and Friday are automatically marked with check marks. On Saturday and Sunday/Holiday the boxes remain empty.
In the Filter range, select Filter time points outside selected interval and select Erase samples under Filter value. This will delete all time points outside the selected interval – Saturdays, Sundays, and public holidays according to the selected public holiday calendar.
The result time series consists only of the remaining weekdays and is not equidistant.
