io.jenetics.prog.regression

Interface LossFunction<T>

Functional Interface:

This is a functional interface and can therefore be used as the assignment target for a lambda expression or method reference.

@FunctionalInterface
public interface LossFunction<T>

This function evaluates how well an evolved program tree fits the given sample data set. If the predictions are totally off, the loss function will output a higher value. If they’re pretty good, it’ll output a lower number. It is the essential part of the overall Error function.

final Error<Double> error = Error.of(LossFunction::mse);

Since:

5.0

Version:

5.0

Author:

Franz Wilhelmstötter

See Also:

Loss function

Method Summary

Modifier and Type	Method and Description
double	apply(T[] calculated, T[] expected) Calculates the error between the expected function values and the values calculated by the actual ProgramGene.
static double	mae(Double[] calculated, Double[] expected) Mean absolute error is measured as the average of sum of absolute differences between predictions and actual observations.
static double	mse(Double[] calculated, Double[] expected) Mean square error is measured as the average of squared difference between predictions and actual observations.
static double	rmse(Double[] calculated, Double[] expected) Root mean square error is measured as the average of squared difference between predictions and actual observations.

Method Detail

apply

double apply(T[] calculated,
             T[] expected)

Calculates the error between the expected function values and the values calculated by the actual ProgramGene.

Parameters:

calculated - the currently calculated function value

expected - the expected function values

Returns:

the error value

Throws:

IllegalArgumentException - if the length of the two arrays are not equal

NullPointerException - if one of the double[] arrays is null

mse

static double mse(Double[] calculated,
                  Double[] expected)

Mean square error is measured as the average of squared difference between predictions and actual observations.

Parameters:

calculated - the function values calculated with the current program tree

expected - the expected function value as given by the sample points

Returns:

the mean square error

Throws:

IllegalArgumentException - if the length of the two arrays are not equal

NullPointerException - if one of the double[] arrays is null

See Also:

rmse(Double[], Double[])

rmse

static double rmse(Double[] calculated,
                   Double[] expected)

Root mean square error is measured as the average of squared difference between predictions and actual observations.

Parameters:

calculated - the function values calculated with the current program tree

expected - the expected function value as given by the sample points

Returns:

the mean square error

Throws:

IllegalArgumentException - if the length of the two arrays are not equal

NullPointerException - if one of the double[] arrays is null

See Also:

mse(Double[], Double[])

mae

static double mae(Double[] calculated,
                  Double[] expected)

Mean absolute error is measured as the average of sum of absolute differences between predictions and actual observations.

Parameters:

calculated - the function values calculated with the current program tree

expected - the expected function value as given by the sample points

Returns:

the mean absolute error

Throws:

IllegalArgumentException - if the length of the two arrays are not equal

NullPointerException - if one of the double[] arrays is null