shark::NearestNeighborModel< InputType, LabelType > Class Template Reference

NearestNeighbor model for classification and regression. More...

#include <shark/Models/NearestNeighborModel.h>

Inheritance diagram for shark::NearestNeighborModel< InputType, LabelType >:

Public Types
enum	DistanceWeights { UNIFORM , ONE_OVER_DISTANCE }
	Type of distance-based weights. More...
typedef AbstractNearestNeighbors< InputType, LabelType >	NearestNeighbors

Public Member Functions
	NearestNeighborModel (NearestNeighbors const *algorithm, unsigned int neighbors=3)
	Constructor.
std::string	name () const
	From INameable: return the class name.
DistanceWeights	getDistanceWeightType () const
	query the way distances enter as weights
void	setDistanceWeightType (DistanceWeights dw)
	set the way distances enter as weights

Detailed Description

template<class InputType, class LabelType>
class shark::NearestNeighborModel< InputType, LabelType >

NearestNeighbor model for classification and regression.

The classification, the model predicts a class label according to a local majority decision among its k nearest neighbors. It is not specified how ties are broken.

For Regression, the (weighted) mean of the k nearest neighbours is computed.

Definition at line 194 of file NearestNeighborModel.h.

Member Typedef Documentation

NearestNeighbors

template<class InputType , class LabelType >

typedef AbstractNearestNeighbors<InputType,LabelType> shark::NearestNeighborModel< InputType, LabelType >::NearestNeighbors

Definition at line 197 of file NearestNeighborModel.h.

Member Enumeration Documentation

DistanceWeights

template<class InputType , class LabelType >

enum shark::NearestNeighborModel::DistanceWeights

Type of distance-based weights.

Enumerator
UNIFORM	uniform (= no) distance-based weights
ONE_OVER_DISTANCE	weight each neighbor's label with 1/distance

Definition at line 201 of file NearestNeighborModel.h.

Constructor & Destructor Documentation

NearestNeighborModel()

template<class InputType , class LabelType >

shark::NearestNeighborModel< InputType, LabelType >::NearestNeighborModel ( NearestNeighbors const *  algorithm, unsigned int  neighbors = 3  )

inline

Constructor.

Parameters

algorithm	the used algorithm for nearest neighbor search
neighbors	number of neighbors

Definition at line 210 of file NearestNeighborModel.h.

Member Function Documentation

getDistanceWeightType()

template<class InputType , class LabelType >

DistanceWeights shark::NearestNeighborModel< InputType, LabelType >::getDistanceWeightType ( ) const

inline

query the way distances enter as weights

Definition at line 219 of file NearestNeighborModel.h.

References shark::NearestNeighborModel< InputType, LabelType >::ONE_OVER_DISTANCE, and shark::NearestNeighborModel< InputType, LabelType >::UNIFORM.

name()

template<class InputType , class LabelType >

std::string shark::NearestNeighborModel< InputType, LabelType >::name ( ) const

inlinevirtual

From INameable: return the class name.

Reimplemented from shark::INameable.

Definition at line 215 of file NearestNeighborModel.h.

setDistanceWeightType()

template<class InputType , class LabelType >

void shark::NearestNeighborModel< InputType, LabelType >::setDistanceWeightType ( DistanceWeights  dw )

inline

set the way distances enter as weights

Definition at line 224 of file NearestNeighborModel.h.

References shark::NearestNeighborModel< InputType, LabelType >::UNIFORM.

---

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

• include/shark/Models/NearestNeighborModel.h