jsat.distributions

Class DistributionSearch

java.lang.Object

jsat.distributions.DistributionSearch

public class DistributionSearch
extends java.lang.Object

Provides methods for selecting the distribution that best fits a given data set.

Constructor Summary

Constructors

Constructor and Description
DistributionSearch()

Method Summary

Modifier and Type	Method and Description
static ContinuousDistribution	getBestDistribution(Vec v) Searches the distributions that are known for a possible fit, and returns what appears to be the best fit.
static ContinuousDistribution	getBestDistribution(Vec v, ContinuousDistribution... possibleDistributions) Searches the distributions that are given for a possible fit, and returns what appears to be the best fit.
static ContinuousDistribution	getBestDistribution(Vec v, double KDECutOff) Searches the distributions that are known for a possible fit, and returns what appears to be the best fit.
static ContinuousDistribution	getBestDistribution(Vec v, double KDECutOff, ContinuousDistribution... possibleDistributions) Searches the distributions that are given for a possible fit, and returns what appears to be the best fit.

Methods inherited from class java.lang.Object

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

Constructor Detail

DistributionSearch

public DistributionSearch()

Method Detail

getBestDistribution

public static ContinuousDistribution getBestDistribution(Vec v)

Searches the distributions that are known for a possible fit, and returns what appears to be the best fit.

Parameters:

v - all the values from a sample

Returns:

the distribution that provides the best fit to the data that this method could find.

getBestDistribution

public static ContinuousDistribution getBestDistribution(Vec v,
                                                         double KDECutOff)

Searches the distributions that are known for a possible fit, and returns what appears to be the best fit. If no suitable fit can be found, a KernelDensityEstimator is fit to the data.

Parameters:

v - all the values from a sample

KDECutOff - the cut off value used for using the KDE. Should be in the range (0, 1). Values less than zero means the KDE will never be used, and greater then 1 means the KDE will always be used.

Returns:

the distribution that provides the best fit to the data that this method could find.

getBestDistribution

public static ContinuousDistribution getBestDistribution(Vec v,
                                                         ContinuousDistribution... possibleDistributions)

Searches the distributions that are given for a possible fit, and returns what appears to be the best fit.

Parameters:

v - all the values from a sample

possibleDistributions - the array of distribution to try and fit to the data

Returns:

the distribution that provides the best fit to the data that this method could find.

getBestDistribution

public static ContinuousDistribution getBestDistribution(Vec v,
                                                         double KDECutOff,
                                                         ContinuousDistribution... possibleDistributions)

Searches the distributions that are given for a possible fit, and returns what appears to be the best fit. If no suitable fit can be found, a KernelDensityEstimator is fit to the data.

Parameters:

v - all the values from a sample

KDECutOff - the cut off value used for using the KDE. Should be in the range (0, 1). Values less than zero means the KDE will never be used, and greater then 1 means the KDE will always be used.

possibleDistributions - the array of distribution to try and fit to the data

Returns:

the distribution that provides the best fit to the data that this method could find.