prtRvGmm - Gaussian Mixture Model Random Variable
RV = prtRvGmm creates a prtRvGmm object with empty
mixingProportions and prtRvMvn components. These parameters can be
set manually or by calling the MLE method. A prtRvGmm is a mixture
of multi-variance normal random variables.
RV = prtRvGmm(PROPERTY1, VALUE1,...) creates a prtRvGmm
object RV with properties as specified by PROPERTY/VALUE pairs.
A prtRvGmm object inherits all properties from the prtRv class.
In addition, it has the following properties:
nComponents - A positive integer specifiying the number of
MVN components in the mixture.
covarianceStructure - The covariance structure applied to each of
the prtRvMvn objects in the mixture. See prtRvMvn.
covariancePool - A logical specifying whether the components
should share a common covariance. If set to
true the covariance of the components are
set to the weighted average of the maximum
likelihood estimate for the covariance
matrices for the components.
components - An array of prtRvMvn objects.
mixingProportions - A discrete probability vector, representing
the probability of each component in the
mixture.
A prtRvGmm object inherits all methods from the prtRv class.
The MLE method can be used to estimate the distribution parameters
from data.
Examples:
ds = prtDataGenOldFaithful; % Load a data set
rv = prtRvGmm('nComponents',2); % Specify 2 components
rv = mle(rv,ds); % Compute the ML estimate
plotPdf(rv); % Plot the estimated PDF
hold on;
plot(ds); % Overlay the original data