R/Zagg.R

In zoevanhavre/Zmix: Overfitted Gaussian mixture models with an unknown number of components.

#' Processing output of label switching function
#'
#' explain here
#' @param USout: output of Zswitch
#' @param .Y: Y, vector of original data
#' @keywords postprocessing
#' @export
#' @examples
#' #nope


Zagg<-	function(USout, .Y=Y){
	#Parameters
	.par<-melt(USout$Pars, id.vars=c("Iteration", "k"))
	theta<-aggregate( value~variable+factor(k), mean ,data=.par)
 	K<-max(.par$k)			
	#Pred
	maxZ<-function (x)  as.numeric(names(which.max(table( x ))))
	Zhat<- factor( apply(t(USout$Zs), 2,maxZ))
	Zemu<-as.numeric(Zhat)
	.Mus<-theta$value[theta$variable=="Mu"]
	for ( i in 1:length(Zemu)){
		Zemu[i]<-.Mus[as.numeric(Zhat[i])]}
	 MSE<-sum((.Y-Zemu)^2)
	 MAE<-sum(abs(.Y-Zemu))

			list("theta"=theta,  "Zpred"=Zhat, "MSE"=MSE, "MAE"=MAE)
			}