...
- outputVariable
Description
Equidistant Timeseries
This transformations calculates the mean value of instantaneous values over a certain period. If the option allowMissingValues is true (this is the default behaviour), then a missing value is returned if one of the input values in the period is a missing value. If the option allowMissingValues is false, then a mean value is calculated if there are 1 or more non-missing values in the aggregation period, i.e. missing values are ignored in this case.
...
Date/Time | Input Value | Calculation | Output Value |
|---|---|---|---|
2007-01-01 00:00 | 1 |
|
|
2007-01-01 06:00 | 2 |
|
|
2007-01-01 12:00 | 3 |
|
|
2007-01-01 18:00 | 4 |
|
|
2007-01-02 00:00 | 5 | (2 + 3 + 4 + 5)/4 | 3,50 |
2007-01-02 06:00 | 6 |
|
|
2007-01-02 12:00 | 7 |
|
|
2007-01-02 18:00 | NaN |
|
|
2007-01-03 00:00 | 9 | - | NaN |
2007-01-03 06:00 | 10 |
|
|
Non Equidistant Timeseries
In the case of a non equidistant timeries, the allowMissingValues option does not bear much meaning, as it is unknown how much (if any) points should be available in the aggregation period. Setting this option does handle the case where a (non equidistant) timestep is available without an actual value (NaN). Furthermore, for non equidistant timeseries the includeFirstValueOfAggregationPeriodInCalculation option is not taken into account, as it is unknown what the first value of the aggregation period would be. The following calculation methodology is executed to determine the mean over the aggregation period:
First the (linearly interpolated) values at the aggregation period boundaries are determined from the non equidistant timeseries. Then, the average of each pair of connected values (including the calculated boundary values) is calculated and multiplied with the respective time between the measurement points. The weighted average of these mean values constitutes the aggregation period mean.
It should be noted that a mean value will be calculated even if no measurement points are available within the aggregation period. This is because the aggregation period boundary values will be calculated when measurement values are available at each side of the aggregation period and these will then be taken as input for the mean value calculation.
The image below visually explains the calculation methodology for non equidistant timeseries, where the interpolated value of one boundary is determined, before the weighted average of the connected pairs is taken. 
Configuration example
...