slope::Score Class Reference

Base class for scoring metrics used in regularized generalized linear regression. More...

#include <score.h>

Inheritance diagram for slope::Score:

Public Member Functions
virtual bool	isWorse (double other, double current) const =0
virtual double	initValue () const =0
std::function< bool(double, double)>	getComparator () const
virtual double	eval (const Eigen::MatrixXd &eta, const Eigen::MatrixXd &y, const std::unique_ptr< Loss > &loss) const =0
virtual double	eval (const Eigen::MatrixXd &eta, const Eigen::MatrixXd &y, const std::unique_ptr< Loss > &loss, const Eigen::VectorXd &observation_weights) const

Static Public Member Functions
static std::unique_ptr< Score >	create (const std::string &metric)

Detailed Description

Base class for scoring metrics used in regularized generalized linear regression.

This abstract class defines the interface for computing various performance metrics that evaluate model predictions against true responses. Derived classes implement specific metrics like MSE, MAE, AUC-ROC etc.

Definition at line 64 of file score.h.

Member Function Documentation

create()

std::unique_ptr< Score > slope::Score::create ( const std::string & metric )

static

Factory method to create specific Score implementations.

Parameters

metric

Name of the scoring metric to create

Returns

Unique pointer to created Score object

eval() [1/2]

virtual double slope::Score::eval ( const Eigen::MatrixXd & eta, const Eigen::MatrixXd & y, const std::unique_ptr< Loss > & loss ) const

pure virtual

Evaluates the scoring metric given predictions and true responses.

Parameters

eta	Matrix of model predictions
y	Matrix of true responses
loss	Loss function used in the model

Returns

Computed score value

Implemented in slope::Accuracy, slope::AUC, slope::Deviance, slope::MAE, slope::MisClass, and slope::MSE.

eval() [2/2]

virtual double slope::Score::eval ( const Eigen::MatrixXd & eta, const Eigen::MatrixXd & y, const std::unique_ptr< Loss > & loss, const Eigen::VectorXd & observation_weights ) const

virtual

Evaluates the metric with observation weights. Empty weights select the unweighted metric. Weights must be finite and nonnegative, and their sum must be positive. Multiplying every weight by a positive constant leaves the score unchanged. Custom metrics can override this overload; the default implementation accepts uniform weights only.

Reimplemented in slope::Accuracy, slope::AUC, slope::Deviance, slope::MAE, slope::MisClass, and slope::MSE.

getComparator()

std::function< bool(double, double)> slope::Score::getComparator

(

)

const

Returns a comparator function for comparing score values.

Returns

Function object that compares two score values

initValue()

virtual double slope::Score::initValue ( ) const

pure virtual

Returns the initial/default value for this scoring metric.

Returns

Initial score value

Implemented in slope::MaximizeScore, and slope::MinimizeScore.

isWorse()

virtual bool slope::Score::isWorse ( double other, double current ) const

pure virtual

Determines if a score value indicates worse performance.

Parameters

other	The score to compare against
current	The current score

Returns

true if other score is worse than current

Implemented in slope::MaximizeScore, and slope::MinimizeScore.

---

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

• include/slope/score.h