TanBayes.default <- function (x, grouping, data, prior = NULL, usekernel = FALSE, fL = 0, ...)
{
    x <- data.frame(x)
    if(!is.factor(grouping))
        stop("grouping/classes object must be a factor")
    if (is.null(prior))
        apriori <- table(grouping) / length(grouping)
    else
        apriori <- as.table(prior / sum(prior))
    call <- match.call()
    Yname <- "grouping"

	mutdata <- discretize(x)
	disdata <- discretize(data)
    mii <- mutinformation(mutdata)

    graph <- graph.adjacency(-mii, mode="directed", weighted=TRUE)
    graph <- minimum.spanning.tree(graph)
    edge_attr(graph, "weight") <- -edge_attr(graph, "weight")
	plot(graph, edge.label=E(graph)$weight)
	root <- which(degree(graph, v = V(graph), mode = "in") == 0, useNames = T)

	tables <- 0
	pstwo <- c()

	est <- function(u) {
		group <- c()
		for(m in 1:nrow(disdata)) {
			if(isTRUE(compare(u, unlist(disdata[m,colselect])))) {
				group <- c(group, mutdata[m, v])
			}
			if(length(group) > 1)
				print(group)
		}

		#print(class(u))
		#print(class(unlist(disdata[1,cls])))
		#group <- c()
		#for(m in 1:nrow(disdata)) {
		#	print(isTRUE(compare(u, disdata[m,cls])))
		#}



		#comparison <- compareEqual(u, disdata[,cls])
		#print(comparison$result)
		#for(m in 1:nrow(disdata)) {
		#	comparison <- compare(u, disdata[m,colselect])
		#	print(comparison$tM)
		#}
		#comparison <-

		#for(m in 1:nrow(disdata)) {
		#	print(u)
			#print(disdata[m,colselect])
		#	print(isTRUE(compare(u, disdata[m,colselect])))

		#}
			#if(isTRUE(compare(u, disdata[m,colselect])))
			#	print("TRUE")
				#group <- c(group, mutdata[m, v])

		#if(length(group) > 1) {
			#pstwo <- pstwo + log(dnorm(mutdata[m, v], mean(group), sd(group)))
			#print(cbind(mean(group), sd(group)))
			#tables <- c(tables, cbind(mean(group), sd(group)))
		#	print(group)
		#}
	}



	for (v in V(graph)) {
		path <- shortest_paths(graph, root, v)$vpath[[1]]
		colselect <- c(ncol(disdata))
		group <- c()

		if(v != root)
			for (p in path)
				colselect <- c(colselect, p)
		else
			colselect <- c(colselect, as.numeric(root))

		colselect <- head(colselect, -1)
		uni <- unique(disdata[,colselect])

		if(is.vector(uni))
			uni <- t(uni)

		apply(uni, 1, est)

		print("#########")
	}

    structure(list(apriori = apriori, tables = tables, levels = levels(grouping),
        call = call, x = x, usekernel = usekernel, varnames = colnames(x)),
        class = "TANBayes")
}