Extrapolate Pattern With Smoothing

Input
Options
Output
Description

Extrapolates by repeating the values in the matching period. The matching period ends at the last non-missing value in the view period. The values in the matching period are referred to as the pattern. This pattern is repeated for the length of the view period. Smoothing is applied on each repetition of the pattern by adding (part of) the difference between the first and last value of the pattern. This linearly declines to zero for the length of the smoothingPeriod.

the "matchingPeriod" option is the period leading up to the last non-missing value. The values in the period are used for the pattern that is repeated.

the "smoothingPeriod" option is the length over which the smoothing is applied. In general, this period should be shorter then the matchingPeriod.

Example: matchingPeriod = 8, Pattern = [4,5,5,5,8,6,7,8], smoothingPeriod = 4 would result in [8,8,7,6,8,6,7,8]

Configuration example

	<transformation id="QO_Extrapolation">
		<interpolationSerial>
			<extrapolatePatternWithSmoothing>
				<inputVariable>
					<timeSeriesSet>
						<moduleInstanceId>QRS_Pre</moduleInstanceId>
						<valueType>scalar</valueType>
						<parameterId>QO</parameterId>
						<locationSetId>DAMS</locationSetId>
						<timeSeriesType>simulated forecasting</timeSeriesType>
						<timeStep unit="hour" multiplier="1"/>
						<relativeViewPeriod unit="hour" start="-24" end="0"/>
						<readWriteMode>add originals</readWriteMode>
					</timeSeriesSet>
				</inputVariable>
				<matchingPeriod unit="hour" multiplier="24"/>
				<smoothingPeriod unit="hour" multiplier="6"/>
				<outputVariable>
					<variableId>GHT_Ex</variableId>
				</outputVariable>
			</extrapolatePatternWithSmoothing>
		</interpolationSerial>
	</transformation>