org.apache.ignite.ml.clustering.gmm

Class GmmModel

java.lang.Object

org.apache.ignite.ml.math.stat.DistributionMixture<MultivariateGaussianDistribution>

org.apache.ignite.ml.clustering.gmm.GmmModel

All Implemented Interfaces:

Serializable, AutoCloseable, DeployableObject, IgniteModel<Vector,Double>, Model<Vector,Double>, Distribution

public class GmmModel
extends DistributionMixture<MultivariateGaussianDistribution>
implements IgniteModel<Vector,Double>, DeployableObject

Gaussian Mixture Model. This algorithm represents a soft clustering model where each cluster is gaussian distribution with own mean value and covariation matrix. Such model can predict cluster using maximum likelihood priciple (see predict(Vector)). Also * this model can estimate probability of given vector (see DistributionMixture.prob(Vector)) and compute likelihood vector where each component of it is a probability of cluster of mixture (see DistributionMixture.likelihood(Vector)).

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
GmmModel(Vector componentProbs, List<MultivariateGaussianDistribution> distributions) Creates an instance of GmmModel.

Method Summary

Modifier and Type	Method and Description
List<Object>	getDependencies() Returns dependencies of this object that can be object with class defined by client side and unknown for server.
Double	predict(Vector input) Make a prediction for the specified input arguments.

Methods inherited from class org.apache.ignite.ml.math.stat.DistributionMixture

componentsProbs, countOfComponents, dimension, distributions, likelihood, prob

Methods inherited from class java.lang.Object

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

Methods inherited from interface org.apache.ignite.ml.IgniteModel

andBefore, andThen, andThen, close, combine, toString

Constructor Detail

GmmModel

public GmmModel(Vector componentProbs,
                List<MultivariateGaussianDistribution> distributions)

Creates an instance of GmmModel.

Parameters:

componentProbs - Probabilities of components.

distributions - Gaussian distributions for each component.

Method Detail

predict

public Double predict(Vector input)

Make a prediction for the specified input arguments.

Specified by:

predict in interface Model<Vector,Double>

Parameters:

input - Input arguments.

Returns:

Prediction result.

getDependencies

public List<Object> getDependencies()

Returns dependencies of this object that can be object with class defined by client side and unknown for server.

Specified by:

getDependencies in interface DeployableObject