Input consists of time series with value properties. Based on those properties statistics are calculated.
- forecastPeak: Calculates 3 statistics, how many peaks are predicted correctly, for how many peak the height is predicted incorrectly and how many peaks are missed. For the calculation input forecast time series with value properties are compared to observed time series
- thresholdExceedanceCount: Calculates 4 statistics, how many threshold exceedences are: forecast but not observed, observed but not forecast, forecast but not within an acceptable time range and for how many are correctly forecast. For the calculation input forecast time series with value properties are compared to observed time series
- forecastOnTimeCount: Count how many forecasts where issued on time. It uses date time properties that specify when the next forecast was supposed to be issued and when it is actually issued. Needs consecutive forecasts because one contains information about the next, so will always need 1 forecast extra before the forecast search period.
- valueCount: Count how many time steps there are in the input time series which contain all value properties with the exact values as configured