weka.estimators

Class UnivariateMixtureEstimator.MM

java.lang.Object

weka.estimators.UnivariateMixtureEstimator.MM

Enclosing class:

UnivariateMixtureEstimator

public class UnivariateMixtureEstimator.MM
extends java.lang.Object

Fast univariate mixture model implementation.

Field Summary

Fields

Modifier and Type	Field and Description
protected int	m_K The number of actual components
protected double[]	m_LogPriors The priors, on log scale
protected double[]	m_Means Means
protected double[]	m_StdDevs Standard deviations

Constructor Summary

Constructors

Constructor and Description
MM()

Method Summary

Modifier and Type	Method and Description
protected void	estimateParameters(double[] values, double[] weights, double[][] probs) Estimate parameters.
void	initializeModel(int K, double[] values, double[] weights, java.util.Random r) Initializes the model.
double	logDensity(double value) Computes log of density for given value.
protected double[]	logJointDensities(double value) Joint densities per cluster.
double	loglikelihood(double[] values, double[] weights) Computes loglikelihood of current model.
protected double	logNormalDens(double x, double mean, double stdDev) Density function of normal distribution.
double	MSE() Returns average of squared errors for current model.
protected int	nearestMean(double val) Returns the index of the nearest mean.
double[][]	predictIntervals(double conf) Returns the interval for the given confidence value.
double	predictQuantile(double percentage) Returns the quantile for the given percentage.
protected double	smallestDistance(double val) Returns smallest distance to given elements.
java.lang.String	toString() Returns string describing the estimator.

Methods inherited from class java.lang.Object

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

Field Detail

m_Means

protected double[] m_Means

Means

m_StdDevs

protected double[] m_StdDevs

Standard deviations

m_LogPriors

protected double[] m_LogPriors

The priors, on log scale

m_K

protected int m_K

The number of actual components

Constructor Detail

MM

public MM()

Method Detail

toString

public java.lang.String toString()

Returns string describing the estimator.

Overrides:

toString in class java.lang.Object

smallestDistance

protected double smallestDistance(double val)

Returns smallest distance to given elements. Assumes m_Means has at least one element (i.e. m_K >= 1).

nearestMean

protected int nearestMean(double val)

Returns the index of the nearest mean. Assumes m_Means has at least one element (i.e. m_K >= 1).

initializeModel

public void initializeModel(int K,
                            double[] values,
                            double[] weights,
                            java.util.Random r)

Initializes the model. Assumes K >= 1, values.length >= 1, and values.length = weights.length.

estimateParameters

protected void estimateParameters(double[] values,
                                  double[] weights,
                                  double[][] probs)

Estimate parameters.

loglikelihood

public double loglikelihood(double[] values,
                            double[] weights)

Computes loglikelihood of current model.

MSE

public double MSE()

Returns average of squared errors for current model.

logNormalDens

protected double logNormalDens(double x,
                               double mean,
                               double stdDev)

Density function of normal distribution.

logJointDensities

protected double[] logJointDensities(double value)

Joint densities per cluster.

logDensity

public double logDensity(double value)

Computes log of density for given value.

predictIntervals

public double[][] predictIntervals(double conf)

Returns the interval for the given confidence value.

Parameters:

conf - the confidence value in the interval [0, 1]

Returns:

the interval

predictQuantile

public double predictQuantile(double percentage)

Returns the quantile for the given percentage.

Parameters:

percentage - the percentage

Returns:

the quantile