org.apache.ignite.ml.composition.boosting.loss

Interface Loss

All Superinterfaces:

Serializable

All Known Implementing Classes:

LogLoss, SquaredError

public interface Loss
extends Serializable

Loss interface of computing error or gradient of error on specific row in dataset.

Method Summary

Modifier and Type	Method and Description
double	error(long sampleSize, double lb, double mdlAnswer) Error value for model answer.
double	gradient(long sampleSize, double lb, double mdlAnswer) Error gradient value for model answer.

Method Detail

error

double error(long sampleSize,
             double lb,
             double mdlAnswer)

Error value for model answer.

Parameters:

sampleSize - Sample size.

lb - Label.

mdlAnswer - Model answer.

Returns:

error value.

gradient

double gradient(long sampleSize,
                double lb,
                double mdlAnswer)

Error gradient value for model answer.

Parameters:

sampleSize - Sample size.

lb - Label.

mdlAnswer - Model answer.

Returns:

error value.