jsat.classifiers.knn

Class LWL

java.lang.Object

jsat.classifiers.knn.LWL

All Implemented Interfaces:

java.io.Serializable, java.lang.Cloneable, Classifier, Parameterized, Regressor

public class LWL
extends java.lang.Object
implements Classifier, Regressor, Parameterized

Locally Weighted Learner (LW) is the combined generalized implementation of Locally Weighted Regression (LWR) and Locally Weighted Naive Bayes (LWNB). The concept is simple, prediction begins in a normal NN style. Instead of returning the prediction result as the average / majority of the found neighbors, a classifier is trained to represent the local area that is represented by the neighbors. The prediction result is then based on this locally trained model.
Because NN search is already slow, and increasing the search size increases the search time, it is recommended to use moderate sized values of k with simple models such as NaiveBayes and LinearRegression.
If the learning algorithm used does not support weighted data points, it will be as if the uniform kernel fucntion was used, regardless of whatever kernel function was set.

See:

• Atkeson, C., Moore, A.,&Schaal, S. (1997). Locally Weighted Learning. Artificial intelligence review, 11–73.
• Frank, E., Hall, M.,&Pfahringer, B. (2003). Locally Weighted Naive Bayes. Proceedings of the Conference on Uncertainty in Artificial Intelligence (pp. 249–256). Morgan Kaufmann.

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
LWL(Classifier classifier, int k, DistanceMetric dm) Creates a new LWL classifier
LWL(Classifier classifier, int k, DistanceMetric dm, KernelFunction kf) Creates a new LWL classifier
LWL(Classifier classifier, int k, DistanceMetric dm, KernelFunction kf, VectorCollection<VecPaired<Vec,java.lang.Double>> vcf) Creates a new LWL classifier
LWL(Regressor regressor, int k, DistanceMetric dm) Creates a new LWL Regressor
LWL(Regressor regressor, int k, DistanceMetric dm, KernelFunction kf) Creates a new LWL Regressor
LWL(Regressor regressor, int k, DistanceMetric dm, KernelFunction kf, VectorCollection<VecPaired<Vec,java.lang.Double>> vcf) Creates a new LWL Regressor

Method Summary

Modifier and Type	Method and Description
CategoricalResults	classify(DataPoint data) Performs classification on the given data point.
LWL	clone()
DistanceMetric	getDistanceMetric() Returns the distance metric in use
KernelFunction	getKernelFunction() Returns the kernel function that will be used to set the weights.
int	getNeighbors() Returns the number of neighbors that will be used to create each local model
static Distribution	guessNeighbors(DataSet d) Guesses the distribution to use for the number of neighbors to consider
double	regress(DataPoint data)
void	setDistanceMetric(DistanceMetric dm) Sets the distance metric that will be used for the nearest neighbor search
void	setKernelFunction(KernelFunction kf) Sets the kernel function that will be used to set the weights of each data point in the local set
void	setNeighbors(int k) Sets the number of neighbors that will be used to create the local model
boolean	supportsWeightedData() Indicates whether the model knows how to train using weighted data points.
void	train(ClassificationDataSet dataSet, boolean parallel) Trains the classifier and constructs a model for classification using the given data set.
void	train(RegressionDataSet dataSet, boolean parallel)

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Methods inherited from interface jsat.classifiers.Classifier

train

Methods inherited from interface jsat.regression.Regressor

train

Methods inherited from interface jsat.parameters.Parameterized

getParameter, getParameters

Constructor Detail

LWL

public LWL(Classifier classifier,
           int k,
           DistanceMetric dm)

Creates a new LWL classifier

Parameters:

classifier - the local classifier to

k - the number of neighbors to create a local classifier from

dm - the metric to use when selecting the nearest points to a query

LWL

public LWL(Classifier classifier,
           int k,
           DistanceMetric dm,
           KernelFunction kf)

Creates a new LWL classifier

Parameters:

classifier - the local classifier to

k - the number of neighbors to create a local classifier from

dm - the metric to use when selecting the nearest points to a query

kf - the kernel function used to weight the local points

LWL

public LWL(Classifier classifier,
           int k,
           DistanceMetric dm,
           KernelFunction kf,
           VectorCollection<VecPaired<Vec,java.lang.Double>> vcf)

Creates a new LWL classifier

Parameters:

classifier - the local classifier to

k - the number of neighbors to create a local classifier from

dm - the metric to use when selecting the nearest points to a query

kf - the kernel function used to weight the local points

vcf - the factory to create vector collections for storing the points

LWL

public LWL(Regressor regressor,
           int k,
           DistanceMetric dm)

Creates a new LWL Regressor

Parameters:

regressor - the local regressor

k - the number of neighbors to create a local classifier from

dm - the metric to use when selecting the nearest points to a query

LWL

public LWL(Regressor regressor,
           int k,
           DistanceMetric dm,
           KernelFunction kf)

Creates a new LWL Regressor

Parameters:

regressor - the local regressor

k - the number of neighbors to create a local classifier from

dm - the metric to use when selecting the nearest points to a query

kf - the kernel function used to weight the local points

LWL

public LWL(Regressor regressor,
           int k,
           DistanceMetric dm,
           KernelFunction kf,
           VectorCollection<VecPaired<Vec,java.lang.Double>> vcf)

Creates a new LWL Regressor

Parameters:

regressor - the local regressor

k - the number of neighbors to create a local classifier from

dm - the metric to use when selecting the nearest points to a query

kf - the kernel function used to weight the local points

vcf - the factory to create vector collections for storing the points

Method Detail

classify

public CategoricalResults classify(DataPoint data)

Description copied from interface: Classifier

Performs classification on the given data point.

Specified by:

classify in interface Classifier

Parameters:

data - the data point to classify

Returns:

the results of the classification.

train

public void train(ClassificationDataSet dataSet,
                  boolean parallel)

Description copied from interface: Classifier

Trains the classifier and constructs a model for classification using the given data set. If the training method knows how, it will used the threadPool to conduct training in parallel. This method will block until the training has completed.

Specified by:

train in interface Classifier

Parameters:

dataSet - the data set to train on

parallel - true if multiple threads should be used to train the model. false if it should be done in a single threaded manner.

supportsWeightedData

public boolean supportsWeightedData()

Description copied from interface: Classifier

Indicates whether the model knows how to train using weighted data points. If it does, the model will train assuming the weights. The values returned by this method may change depending on the parameters set for the model.

Specified by:

supportsWeightedData in interface Classifier

Specified by:

supportsWeightedData in interface Regressor

Returns:

true if the model supports weighted data, false otherwise

regress

public double regress(DataPoint data)

Specified by:

regress in interface Regressor

train

public void train(RegressionDataSet dataSet,
                  boolean parallel)

Specified by:

train in interface Regressor

clone

public LWL clone()

Specified by:

clone in interface Classifier

Specified by:

clone in interface Regressor

Overrides:

clone in class java.lang.Object

setNeighbors

public void setNeighbors(int k)

Sets the number of neighbors that will be used to create the local model

Parameters:

k - the number of neighbors to obtain

getNeighbors

public int getNeighbors()

Returns the number of neighbors that will be used to create each local model

Returns:

the number of neighbors that will be used

setDistanceMetric

public void setDistanceMetric(DistanceMetric dm)

Sets the distance metric that will be used for the nearest neighbor search

Parameters:

dm - the distance metric to use for nearest neighbor search

getDistanceMetric

public DistanceMetric getDistanceMetric()

Returns the distance metric in use

Returns:

the distance metric in use

setKernelFunction

public void setKernelFunction(KernelFunction kf)

Sets the kernel function that will be used to set the weights of each data point in the local set

Parameters:

kf - the kernel function to use for weighting

getKernelFunction

public KernelFunction getKernelFunction()

Returns the kernel function that will be used to set the weights.

Returns:

the kernel function that will be used to set the weights

guessNeighbors

public static Distribution guessNeighbors(DataSet d)

Guesses the distribution to use for the number of neighbors to consider

Parameters:

d - the dataset to get the guess for

Returns:

the guess for the Neighbors parameter