xgboost
Public Member Functions | List of all members
xgboost::common::AFTLoss Class Reference

The AFT loss function. More...

#include <survival_util.h>

Collaboration diagram for xgboost::common::AFTLoss:
Collaboration graph

Public Member Functions

 AFTLoss (ProbabilityDistributionType dist_type)
 Constructor for AFT loss function. More...
 
double Loss (double y_lower, double y_upper, double y_pred, double sigma)
 Compute the AFT loss. More...
 
double Gradient (double y_lower, double y_upper, double y_pred, double sigma)
 Compute the gradient of the AFT loss. More...
 
double Hessian (double y_lower, double y_upper, double y_pred, double sigma)
 Compute the hessian of the AFT loss. More...
 

Detailed Description

The AFT loss function.

Constructor & Destructor Documentation

◆ AFTLoss()

xgboost::common::AFTLoss::AFTLoss ( ProbabilityDistributionType  dist_type)
inlineexplicit

Constructor for AFT loss function.

Parameters
dist_typeChoice of probability distribution for the noise term in AFT

Member Function Documentation

◆ Gradient()

double xgboost::common::AFTLoss::Gradient ( double  y_lower,
double  y_upper,
double  y_pred,
double  sigma 
)

Compute the gradient of the AFT loss.

Parameters
y_lowerLower bound for the true label
y_upperUpper bound for the true label
y_predPredicted label
sigmaScaling factor to be applied to the distribution of the noise term

◆ Hessian()

double xgboost::common::AFTLoss::Hessian ( double  y_lower,
double  y_upper,
double  y_pred,
double  sigma 
)

Compute the hessian of the AFT loss.

Parameters
y_lowerLower bound for the true label
y_upperUpper bound for the true label
y_predPredicted label
sigmaScaling factor to be applied to the distribution of the noise term

◆ Loss()

double xgboost::common::AFTLoss::Loss ( double  y_lower,
double  y_upper,
double  y_pred,
double  sigma 
)

Compute the AFT loss.

Parameters
y_lowerLower bound for the true label
y_upperUpper bound for the true label
y_predPredicted label
sigmaScaling factor to be applied to the distribution of the noise term

The documentation for this class was generated from the following file: