Class SignedBias
java.lang.Object
net.simplace.sim.util.statistics.ErrorCalculationMethod
net.simplace.sim.util.statistics.Bias
net.simplace.sim.util.statistics.SignedBias
Methods for calculating the bias/mean error and comparing signed values.
See D. Wallach, Working with Dynamic Crop Models, p. 22
Calculation of bias is the same as for the Bias class.
Comparison is different. Bias chooses the set with
the smallest absolute value (the closest to 0), whereas
SignedBias chooses the set with the smallest signed
value (the closest to -Infinity).
It can be used to select the maximum of simulated values by
setting observed values to 0: the bias is then the negative mean
of the simulated values, so the best (smallest) SignedBias belongs
to the largest simulated values.
- Author:
- Gunther Krauss
-
Field Summary
Fields inherited from class ErrorCalculationMethod
count, sum -
Constructor Summary
Constructors -
Method Summary
Methods inherited from class Bias
addDataPair, getValueMethods inherited from class ErrorCalculationMethod
getWorstValue, init
-
Constructor Details
-
SignedBias
public SignedBias()
-
-
Method Details
-
isBetter
public boolean isBetter(double better, double worse) Description copied from class:ErrorCalculationMethodChecks if the first value is better than the second one with respect to the calculation method (e.g. smaller for errors, closer to 1 for model efficiency). -
getLabel
-
getDescription
- Overrides:
getDescriptionin classBias- Returns:
- the full name or a short description of the method (empty by default)
-