mml

Class Mixture.Est

java.lang.Object

la.la.Value

la.la.Function

la.la.Function.Native

mml.Estimator

mml.UPModel.Est

mml.Mixture.Est

All Implemented Interfaces:

java.lang.Comparable<Value>

Enclosing class:

Mixture

public class Mixture.Est
extends UPModel.Est

The Unparameterised Mixture's Estimator, capable of estimating a fully parameterised Mixture Model.

Nested Class Summary

Nested classes/interfaces inherited from class la.la.Function.Native

Function.Native.WithInverse

Nested classes/interfaces inherited from class la.la.Function

Function.Cts2Cts, Function.Cts2Cts2Cts, Function.CtsD2CtsD, Function.HasInverse, Function.Native, Function.Native2, Function.Native3

Nested classes/interfaces inherited from class la.la.Value

Value.Atomic, Value.Bool, Value.Char, Value.Chars, Value.Cts, Value.Defer, Value.Discrete, Value.Enum, Value.Inc_Or, Value.Int, Value.Lambda, Value.List, Value.Maybe, Value.Option, Value.Real, Value.Scannable, Value.Structured, Value.Triv, Value.Tuple

Field Summary

Fields

Modifier and Type	Field and Description
Estimator	classEst The estimator for an individual class (cluster, component) of the Mixture Model.

Fields inherited from class mml.Estimator

params

Fields inherited from class la.la.Value

C, comparator, CR, E, eight, ffalse, five, fiveR, four, fourR, half, negCR, negOne, negOneR, negTenR, nilVal, nine, None, one, oneR, PI, PIby2, point01, point1, point5, point9, quarter, seven, six, ten, tenR, three, threeR, triv, ttrue, two, twoR, zero, zeroR

Constructor Summary

Constructors

Constructor and Description
Est(Value ps) Note that classEst = upm.estimator(ps).

Method Summary

Modifier and Type	Method and Description
Mixture.M	EM(double[][] mshp, Vector ds) Find a good Mixture Model starting from the given memberships.
Mixture.M	EM(Mixture.M mx, double[][] mshp, Vector ds) The core Expectation Maximization algorithm.
Mixture.M	EM(Mixture.M mx, Vector ds) Find a good Mixture Model starting from the given Model, mx.
Mixture.M	join(Mixture.M mx, int c1, int c2, Vector ds) Merge classes c1 and c2 into one class, and adjust.
Mixture.M	kill(Mixture.M mx, int c, Vector ds) Remove class 'c', redistribute its members, and adjust.
Mixture.M	makeNclasses(int nClass, Vector ds) Try to Find a good Mixture Model with 'nClass' classes (clusters).
void	Model2mshp(Mixture.M mx, double[][] mshp, Vector ds) Given a Mixture Model, mx, return the (fractional) class memberships, mshp, of the data.
double[][]	Model2mshp(Mixture.M mx, Vector ds) See Model2mshp(mml.Mixture.M, double[][], la.maths.Vector).
double[]	mshp2abndc(double[][] mshp, Vector ds) Calculate class abundances from class memberships, mshp.
Mixture.M	mshp2Model(double[][] mshp, Vector ds) Given (possibly fractional) class memberships, mshp, (re-)estimate the Mixture Model (but note that msg2() is zero).
double[][]	randomMshp(int nClass, int nData) Return a matrix of random class memberships.
Mixture.M	split(Mixture.M mx, int c, Vector ds) Split class 'c', replace it with two sub-classes, and adjust.
Mixture.M	ss2Model(Value ss) Given ss (=ds), estimate a Mixture Model.

Methods inherited from class mml.UPModel.Est

defnParams, sp2Model, stats, stats, toString

Methods inherited from class mml.Estimator

apply, asUPModel, ds2Model, ds2ModelSp, ss2ModelSp, stats, stats

Methods inherited from class la.la.Function

bOp, fromBop, fromUop, main, type, uOp

Methods inherited from class la.la.Value

AoM, bOp, compareTo, cons, cts, elt, errMsg, error, force, isTuple, just, n, nElts, nlAoM, pair, print, quad, real, RTE, triple, tuple, UOE, x

Methods inherited from class java.lang.Object

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

Field Detail

classEst

public final Estimator classEst

The estimator for an individual class (cluster, component) of the Mixture Model.

Constructor Detail

Est

public Est(Value ps)

Note that classEst = upm.estimator(ps).

Method Detail

ss2Model

public Mixture.M ss2Model(Value ss)

Given ss (=ds), estimate a Mixture Model. ss2Model carries out a heuristic search, calling EM, split, kill and join until convergence.

Specified by:

ss2Model in class Estimator

EM

public Mixture.M EM(Mixture.M mx,
                    double[][] mshp,
                    Vector ds)

The core Expectation Maximization algorithm. Alternately estimate memberhips from a Model, and a Model from memberships. To start, 'mx' or 'mshp', but not both, may be null.

EM

public Mixture.M EM(double[][] mshp,
                    Vector ds)

Find a good Mixture Model starting from the given memberships. The number of classes (clusters) is not changed.

EM

public Mixture.M EM(Mixture.M mx,
                    Vector ds)

Find a good Mixture Model starting from the given Model, mx. The number of classes (clusters) is not changed.

makeNclasses

public Mixture.M makeNclasses(int nClass,
                              Vector ds)

Try to Find a good Mixture Model with 'nClass' classes (clusters). It starts from random memberships.

kill

public Mixture.M kill(Mixture.M mx,
                      int c,
                      Vector ds)

Remove class 'c', redistribute its members, and adjust.

split

public Mixture.M split(Mixture.M mx,
                       int c,
                       Vector ds)

Split class 'c', replace it with two sub-classes, and adjust.

join

public Mixture.M join(Mixture.M mx,
                      int c1,
                      int c2,
                      Vector ds)

Merge classes c1 and c2 into one class, and adjust.

mshp2Model

public Mixture.M mshp2Model(double[][] mshp,
                            Vector ds)

Given (possibly fractional) class memberships, mshp, (re-)estimate the Mixture Model (but note that msg2() is zero). Memberships are stored by columns for use as weights.

mshp2abndc

public double[] mshp2abndc(double[][] mshp,
                           Vector ds)

Calculate class abundances from class memberships, mshp. Note, 'ds' is used to provide the data weights.

Model2mshp

public void Model2mshp(Mixture.M mx,
                       double[][] mshp,
                       Vector ds)

Given a Mixture Model, mx, return the (fractional) class memberships, mshp, of the data.

Model2mshp

public double[][] Model2mshp(Mixture.M mx,
                             Vector ds)

See Model2mshp(mml.Mixture.M, double[][], la.maths.Vector). This is another parameterisation.

randomMshp

public double[][] randomMshp(int nClass,
                             int nData)

Return a matrix of random class memberships. It has 'nClass' classes (clusters, columns) and 'nData' rows, stored by columns for use as weights.