library(class)
library(e1071)
source("data")
x <- subset(data, select = -Class)
y <- c(rep(1,sum(data$Class == 1)),rep(2,sum(data$Class == 2)))
pred <- rep(0, length(y))
ng <- length(unique(y))
np <- dim(x)
ncb <-ng * round((0.4*ng*(ng-1 + np[2]/2))/ng,0) #modified version of original suggestion from Kohonen, but here with equal CBVs for all groups
for(iter in 1:length(y))
{
         set.seed(0) # note that LVQ in R is only reproducible given a fixed seed
         mod0 <- lvqinit(x[-iter,], y[-iter], ncb, prior=c(0.5,0.5))
         mod1 <- olvq1(x[-iter,], y[-iter], mod0) 
         mod1 <- lvq3(x[-iter,], y[-iter], mod1)
         pred[iter] <- lvqtest(mod1, x[iter,])
}
cM <- table(true = data$Class, predicted = pred)
cMstats <- classAgreement(cM)

# cat("\n Multipass LVQ \n")
# cat(" ============================== ")
# cat("\n Accuracy: ") ; print(cMstats$diag)
# cat("\n Kappa: ") ; print(cMstats$kappa)
# cat("\n Confusion matrix: \n") ; print(cM)
# cat("\n")

write(cM, 'confusionMatrix')
write(cMstats$diag, 'accuracy')
write(cMstats$kappa, 'kappa')
write(pred, file="pred_lvq", ncolumns = 1)