Class GaussianMixture

java.lang.Object

smile.stat.distribution.Mixture

smile.stat.distribution.ExponentialFamilyMixture

smile.stat.distribution.GaussianMixture

All Implemented Interfaces:

Serializable, Distribution

public class GaussianMixture
extends ExponentialFamilyMixture

Finite univariate Gaussian mixture. The EM algorithm is provided to learn the mixture model from data. BIC score is employed to estimate the number of components.

See Also:

• Serialized Form

Nested Class Summary

Nested classes/interfaces inherited from class Mixture

Mixture.Component

Field Summary

Fields inherited from class ExponentialFamilyMixture

bic, L

Fields inherited from class Mixture

components

Constructor Summary

Constructors

Constructor	Description
GaussianMixture(Mixture.Component... components)	Constructor.

Method Summary

Modifier and Type	Method	Description
static GaussianMixture	fit(double[] x)	Fits the Gaussian mixture model with the EM algorithm.
static GaussianMixture	fit(int k, double[] x)	Fits the Gaussian mixture model with the EM algorithm.

Methods inherited from class ExponentialFamilyMixture

fit, fit

Methods inherited from class Mixture

bic, cdf, entropy, length, logp, map, mean, p, posteriori, quantile, rand, size, toString, variance

Methods inherited from class Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Methods inherited from interface Distribution

inverseTransformSampling, likelihood, logLikelihood, quantile, quantile, rand, rejectionSampling, sd

Constructor Details

GaussianMixture

public GaussianMixture(Mixture.Component... components)

Constructor.

Parameters:

components - a list of multivariate Gaussian distributions.

Method Details

fit

public static GaussianMixture fit(int k,
 double[] x)

Fits the Gaussian mixture model with the EM algorithm.

Parameters:

k - the number of components.

x - the training data.

Returns:

the distribution.

fit

public static GaussianMixture fit(double[] x)

Fits the Gaussian mixture model with the EM algorithm. The number of components will be selected by BIC.

Parameters:

x - the training data.

Returns:

the distribution.