library(e1071)
source("data")
pred <- rep(0, length(data$Class))
# added for use of class.weights
#w1 <- sum(data$Class == 1) / length(data$Class)
#w2 <- sum(data$Class == 2) / length(data$Class)
# w1 <- 1  ; w2 <- 1
# if (sum(data$Class == 1) > sum(data$Class == 2)) 
# {
# 	w1 <- 1 ; w2 <- round(sum(data$Class == 1) / sum(data$Class == 2))
# }
# if (sum(data$Class == 1) < sum(data$Class == 2))
# {
# 	w2 <- 1 ; w1 <- round(sum(data$Class == 2) / sum(data$Class == 1))
# }
for(iter in 1:length(data$Class))
{
	set.seed(iter)
	mod <- svm(Class ~ ., data = data[-iter,], cache=512, kernel="sigmoid") #,
	 #class.weights = c("1"=w1,"2"=w2))
	 # must have equal weights to have entropy near zero with random groups!!!
 	pred[iter] <- predict(mod, data[iter,])
}
cM <- table(true = data$Class, pred = pred)
cMstats <- classAgreement(cM)

# cat("\n SVM with Hyperbolic Tangent kernel function \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_svm", ncolumns = 1)