require(RnavGraph)
require(PairViz)
# require(hanna)
data(olive)
require(RDRToolbox)
require(cluster)
require(mclust)
require(kernlab)
require(vegan)


data <- data.frame(scale(olive[,-(1:2)]))
pointlabels <- as.character(olive[,"Area"])

################################
#
# k-means
#
################################

k <- 9
kmdata <- kmeans(data,k)
groups <- kmdata$cluster

ng.data <- ng_data(name = "OliveKM",
		data = data,
		shortnames = c("p1","p2","s","oleic","l1","l2","a","e"),
	  group = groups,
		labels = as.character(olive[,"Area"])
)

glyphs <- as.vector(apply(hpaths(names(ng.data)),1, FUN = function(x){x}))

G <- completegraph(shortnames(ng.data))
LG <- linegraph(G)
ng.lg <- ng_graph("3d transitions",LG, layout = 'kamadaKawaiSpring')
ng.lgnot <- ng_graph("4d transitions",complement(LG), layout = 'kamadaKawaiSpring')



nav <- navGraph(ng.data,
		list(ng.lg,ng.lgnot),
		list(ng_2d(ng.data,ng.lg,glyphs = glyphs,ng_2d(ng.data,ng.lgnot)))

nav1 <- navGraph(ng.data,
		list(ng.lg,ng.lgnot),
		list(ng_2d(ng.data,ng.lg,glyphs = hpaths(shortnames(ng.data))),ng_2d(ng.data,ng.lgnot)))
	
################################
#
# model based
#
################################

mcdata <- Mclust(data, 1:20)
groups <- mcdata$classification

ng.data <- ng_data(name = "OliveMC",
		data = data,
		shortnames = c("p1","p2","s","oleic","l1","l2","a","e"),
	  group = groups,
		labels = pointlabels
)

LG <- linegraph(G)
ng.lg <- ng_graph("3d transitions",LG, layout = 'kamadaKawaiSpring')
ng.lgnot <- ng_graph("4d transitions",complement(LG), layout = 'kamadaKawaiSpring')

glyphs <- hpaths(shortnames(ng.data))


nav <- navGraph(ng.data,
		list(ng.lg,ng.lgnot),
		list(ng_2d(ng.data,ng.lg,glyphs = glyphs), ng_2d(ng.data,ng.lgnot)))

nav1 <- navGraph(ng.data,
		list(ng.lg,ng.lgnot),
		list(ng_2d(ng.data,ng.lg,glyphs = glyphs), ng_2d(ng.data,ng.lgnot)))

scagNav(data = ng.data,
				 scags = c("Monotonic", "NotMonotonic",
				           "Clumpy", "NotClumpy",
							      "Convex", "NotConvex",  
							      "Stringy", "NotStringy",
							      "Skinny", "NotSkinny",
							      "Outlying","NotOutlying",
							      "Sparse", "NotSparse",
							      "Striated", "NotStriated",
							      "Skewed", "NotSkewed"),
							glyphs = glyphs,
					topFrac = 0.15,
					sep = ":")

x <- runif(nrow(ng.data@data))
y <- runif(nrow(ng.data@data))
handler <- tk2d_init(x = x, y = y, data = ng.data@data, name = ng.data@name, labels = pointlabels, glyphs = hpaths(names(ng.data)))

################################
#
# agnes hierarchical
# Single linkage
#
################################

agdata <- agnes(data, method="single")
groups <- cutree(as.hclust(agdata),h=1.2)

ng.data <- ng_data(name = "OliveAgSingLink",
		data = data,
		shortnames = c("p1","p2","s","oleic","l1","l2","a","e"),
	  group = groups,
		labels = as.character(olive[,"Area"])
)


G <- completegraph(shortnames(ng.data))
LG <- linegraph(G)
ng.lg <- ng_graph("3d transitions",LG, layout = 'kamadaKawaiSpring')
ng.lgnot <- ng_graph("4d transitions",complement(LG), layout = 'kamadaKawaiSpring')



nav <- navGraph(ng.data,
		list(ng.lg,ng.lgnot),
		list(ng_2d(ng.data,ng.lg,glyphs = hpaths(shortnames(ng.data))),ng_2d(ng.data,ng.lgnot)))

nav1 <- navGraph(ng.data,
		list(ng.lg,ng.lgnot),
		list(ng_2d(ng.data,ng.lg,glyphs = hpaths(shortnames(ng.data))),ng_2d(ng.data,ng.lgnot)))
	
################################
#
# agnes hierarchical
# Average linkage
#
################################

agdata <- agnes(data, method="average")
groups <- cutree(as.hclust(agdata),h=3)

ng.data <- ng_data(name = "OliveAgAveLink",
		data = data,
		shortnames = c("p1","p2","s","oleic","l1","l2","a","e"),
	  group = groups,
		labels = as.character(olive[,"Area"])
)

G <- completegraph(shortnames(ng.data))
LG <- linegraph(G)
ng.lg <- ng_graph("3d transitions",LG, layout = 'kamadaKawaiSpring')
ng.lgnot <- ng_graph("4d transitions",complement(LG), layout = 'kamadaKawaiSpring')

nav <- navGraph(ng.data,
		list(ng.lg,ng.lgnot),
		list(ng_2d(ng.data,ng.lg,glyphs = hpaths(shortnames(ng.data))),ng_2d(ng.data,ng.lgnot)))

nav1 <- navGraph(ng.data,
		list(ng.lg,ng.lgnot),
		list(ng_2d(ng.data,ng.lg,glyphs = hpaths(shortnames(ng.data))),ng_2d(ng.data,ng.lgnot)))
	
	
################################
#
# agnes hierarchical
# Complete linkage
#
################################

agdata <- agnes(data, method="complete")
groups <- cutree(as.hclust(agdata),h=4)

ng.data <- ng_data(name = "OliveAgCmpLink2",
		data = data,
		shortnames = c("p1","p2","s","oleic","l1","l2","a","e"),
	  group = groups,
		labels = as.character(olive[,"Area"])
)


G <- completegraph(shortnames(ng.data))
LG <- linegraph(G)
ng.lg <- ng_graph("3d transitions",LG, layout = 'kamadaKawaiSpring')
ng.lgnot <- ng_graph("4d transitions",complement(LG), layout = 'kamadaKawaiSpring')



nav <- navGraph(ng.data,
		list(ng.lg,ng.lgnot),
		list(ng_2d(ng.data,ng.lg,glyphs = hpaths(shortnames(ng.data))),ng_2d(ng.data,ng.lgnot)))

nav1 <- navGraph(ng.data,
		list(ng.lg,ng.lgnot),
		list(ng_2d(ng.data,ng.lg,glyphs = hpaths(shortnames(ng.data))),ng_2d(ng.data,ng.lgnot)))


################################
#
# kernel k-means
#
################################

k <- 9
kkmdata <- kkmeans(as.matrix(data),k)

groups <- kkmdata
kpcs <-kpca(data,features=8)
ng.data <- ng_data(name = "Olive Kernel KM",
		data = kpcs, #data,
		#shortnames = c("p1","p2","s","oleic","l1","l2","a","e"),
	  group = groups,
		labels = as.character(olive[,"Area"])
)


G <- completegraph(shortnames(ng.data))
LG <- linegraph(G)
ng.lg <- ng_graph("3d transitions",LG, layout = 'kamadaKawaiSpring')
ng.lgnot <- ng_graph("4d transitions",complement(LG), layout = 'kamadaKawaiSpring')



nav <- navGraph(ng.data,
		list(ng.lg,ng.lgnot),
		list(ng_2d(ng.data,ng.lg,glyphs = hpaths(shortnames(ng.data))),ng_2d(ng.data,ng.lgnot)))

nav1 <- navGraph(ng.data,
		list(ng.lg,ng.lgnot),
		list(ng_2d(ng.data,ng.lg,glyphs = hpaths(shortnames(ng.data))),ng_2d(ng.data,ng.lgnot)))


##########################
#
#  A box in arbitrary d
#

box_d <- function(nface=15, p=3, cut=0.5, noise=0.0){			n <- nface * 2 * p
      face <- rep(cut, nface)
      if (noise > 0) face <- face + runif(nface,-noise,noise)
      result <- matrix(0,nrow=n,ncol=p)
      for (j in 1:p) {
      	result[,j] <- runif(n,-cut, cut)
      	}
      	
      for (j in 1:p) {
      	start <- (2*(j-1)*nface) + 1
      	end <- 2*j*nface
      	half <- end - nface
      	#rows <- seq(start,half,by=1)
      	#result[start:half,j] <- face   
      	result[(start:half),j] <- face
   	    result[((half+1):end),j] <- - face      	      }

            result
	}

b6 <- box_d(nface=20,p=6,cut=10)
groups <- rep(seq(1, nrow(b6)/20),20)
groups <- matrix(groups,ncol=20, byrow = FALSE)
groups <-  c(groups[1,], groups[2,], groups[3,], groups[4,],
             groups[5,], groups[6,], groups[7,], groups[8,],
             groups[9,], groups[10,], groups[11,], groups[12,])


mcdata <- Mclust(b6, 1:20)
groups <- mcdata$classification


dims <- 4 
dise <- vegdist(b6, method="euclidean")
ord <- isomap(dise,k = 12, ndim= dims, fragmentedOK = TRUE)

			
df.b6 <- data.frame(b6)

ng.data <- ng_data(name = "Six D box raw data",
		data = df.b6,
		#shortnames = c("p1","p2","s","oleic","l1","l2","a","e"),
	  group = groups
)
		
iso.b6 <- data.frame(ord$points)

ng.data <- ng_data(name = "Six D box Isomap",
		data = iso.b6,
		#shortnames = c("p1","p2","s","oleic","l1","l2","a","e"),
	  group = groups
)

lle.b6 <- data.frame(LLE(b6,dim=5,k=5))
ng.data <- ng_data(name = "Six D box LLE",
		data = lle.b6,
		#shortnames = c("p1","p2","s","oleic","l1","l2","a","e"),
	  group = groups
)


G <- completegraph(names(ng.data))
LG <- linegraph(G)
ng.lg <- ng_graph("3d transitions",LG, layout = 'kamadaKawaiSpring')
ng.lgnot <- ng_graph("4d transitions",complement(LG), layout = 'kamadaKawaiSpring')

nav <- navGraph(ng.data,
		list(ng.lg,ng.lgnot),
		list(ng_2d(ng.data,ng.lg,glyphs = hpaths(shortnames(ng.data))),ng_2d(ng.data,ng.lgnot)))

nav1 <- navGraph(ng.data,
		list(ng.lg,ng.lgnot),
		list(ng_2d(ng.data,ng.lg,glyphs = hpaths(shortnames(ng.data))),ng_2d(ng.data,ng.lgnot)))

scagNav(df.b6,
					scags = c("Clumpy", "NotClumpy",
							"Monotonic", "NotMonotonic",
							"Convex", "NotConvex",  
							"Stringy", "NotStringy",
							"Skinny", "NotSkinny",
							"Outlying","NotOutlying",
							"Sparse", "NotSparse",
							"Striated", "NotStriated",
							"Skewed", "NotSkewed"),
					topFrac = 0.05)