Estimate Multivariate Misclassification Error with Gaussian Finite Mixture Model #R #clustering #mul

estimate misclassification error of your multivariate model for machine learning validation

#libraries library(FactoMineR) library(mclust) data(decathlon) #prepare data class<-decathlon$Competition data<-decathlon[,1:12] #Bayesian Information Criterion (BIC) BIC <- mclustBIC(data) plot(BIC) #choice of the model: hard computing! mod <- MclustDA(data, class, modelType = "EDDA") summary(mod)

------------------------------------------------ Gaussian finite mixture model for classification ------------------------------------------------ EDDA model summary: log.likelihood n df BIC -629.6797 41 102 -1638.144 Classes n Model G Decastar 13 EEE 1 OlympicG 28 EEE 1 Training classification summary: Predicted Class Decastar OlympicG Decastar 13 0 OlympicG 0 28 Training error = 0