SDMIXTURE Estimating Gaussian mixture model P=SDMIXTURE(DATA,options) P=SDMIXTURE(DATA,C) INPUT DATA data set object C Number of components (scalar or vector with a number per class) OUTPUT P mixture model pipeline OPTIONS 'comp',C number of components per class (def: 'auto' = choose automatically) 'iter',I number of iterations (def: 100) 'cluster' return clustering result (one output per component) 'reg',R regularize by adding scalar R to covariance diagonals (def: R=0) 'prior',P vector with class priors (default: use priors from the training set) 'maxsamples',N maximum num.of samples used for auto.cluster count estimation (def: 500) 'cluster grid',G define a vector with number of clusters for grid search (def: 1:10) 'init',P initialize EM algorithm with Gaussian pipeline p FIELDS P.complab return labels of mixture components P(1,IND) Make a subset of components given component indices IND DESCRIPTION SDMIXTURE estimate Gaussian mixture model for each class in DATA. By default, the number of mixture components is estimated from data. Number of components may be also provided as second parameter or in the 'comp' option. SDMIXTURE may be used for unsupervised cluster analysis using the 'cluster' option. If specified, SDMIXTURE will perform mixture estimation on each class present and return one output per cluster. READ MORE http://perclass.com/doc/guide/classifiers.html#sdmixture SEE ALSO SDGAUSS, SDPARZEN
sdmixture
is referenced in examples: