I know I can plot a dendrogram as follows
library(cluster)
d <- mtcars
d[,8:11] <- lapply(d[,8:11], as.factor)
gdist <- daisy(d, metric = c("gower"), stand = FALSE)
dendro <- hclust(gdist, method = "average")
plot(as.dendrogram(dendro))
However I have some groups identified (eg. by an iterative classification method), given as the last column in d
G <- c(1,2,3,3,4,4,5,5,5,5,1,2,1,1,2,4,1,3,4,5,1,7,4,3,3,2,1,1,1,3,5,6)
d$Group <- G
head(d)
mpg cyl disp hp drat wt qsec vs am gear carb Group
Mazda RX4 21.0 6 160 110 3.90 2.620 16.46 0 1 4 4 1
Mazda RX4 Wag 21.0 6 160 110 3.90 2.875 17.02 0 1 4 4 2
Datsun 710 22.8 4 108 93 3.85 2.320 18.61 1 1 4 1 3
Hornet 4 Drive 21.4 6 258 110 3.08 3.215 19.44 1 0 3 1 3
Hornet Sportabout 18.7 8 360 175 3.15 3.440 17.02 0 0 3 2 4
Valiant 18.1 6 225 105 2.76 3.460 20.22 1 0 3 1 4
I am trying to plot all the dendrograms together on the same plot with the same scale. The groups with only a single member also needs to be plotted. (group 6 and 7)
I am able to plot individual dendrograms for subset of the data except when number of members in a group is only one. But I don't think this is the right approach.
layout(matrix(1:9, 3,3,byrow=TRUE))
gdist <- as.matrix(gdist)
N <- max(G)
for (i in 1:N){
rc_tokeep <- row.names(subset(d, G==i))
dis <- as.dist(gdist[rc_tokeep, rc_tokeep])
dend <- hclust(dis, method = "average")
plot(as.dendrogram(dend))
}
The loop is giving this error for the last two groups. (6 and 7) having only a single member.
Error in hclust(dis, method = "average") :
must have n >= 2 objects to cluster
Essentially I wan't to reproduce these type of plots. The clusters with single members are also plotted here.
If you want to mimic the last few graphs, you can do something like this:
N <- max(G)
layout(matrix(c(0,1:N,0),nc=1))
gdist <- as.matrix(gdist)
for (i in 1:N){
par(mar=c(0,3,0,7))
rc_tokeep <- row.names(subset(d, G==i))
if(length(rc_tokeep)>2){ #The idea is to catch the groups with one single element to plot them differently
dis <- as.dist(gdist[rc_tokeep, rc_tokeep])
dend <- hclust(dis, method = "average")
plot(as.dendrogram(dend),horiz=TRUE,
xlim=c(.8,0),axes=FALSE) # giving the same xlim will scale all of them, here i used 0.8 to fit your data but you can change it to whatever
}else{
plot(NA,xlim=c(.8,0),ylim=c(0,1),axes=F,ann=F)
segments(0,.5,.1,.5) #I don't know how you intend to compute the length of the branch in a group of 1 element, you might want to change that
text(0,.5, pos=4,rc_tokeep,xpd=TRUE)
}
}
With your example it gives:
If you want to add the scale you can add a grid in all graphs and a scale in the last one:
N <- max(G)
layout(matrix(c(0,1:N,0),nc=1))
gdist <- as.matrix(gdist)
for (i in 1:N){
par(mar=c(0,3,0,7))
rc_tokeep <- row.names(subset(d, G==i))
if(length(rc_tokeep)>2){
dis <- as.dist(gdist[rc_tokeep, rc_tokeep])
dend <- hclust(dis, method = "average")
plot(as.dendrogram(dend),horiz=TRUE,xlim=c(.8,0),xaxt="n",yaxt="n")
abline(v=seq(0,.8,.1),lty=3) #Here the grid
}else{
plot(NA,xlim=c(.8,0),ylim=c(0,1),axes=F,ann=F)
segments(0,.5,.1,.5)
text(0,.5, pos=4,rc_tokeep,xpd=TRUE)
abline(v=seq(0,.8,.1),lty=3) #Here the grid
}
}
axis(1,at=seq(0,.8,.1)) #Here the axis
And finally if you want to even the spaces between the different branches in the resulting plot, you can use table(d$Group)
to get the number of members of each group and use it as a height for each subplot:
N <- max(G)
layout(matrix(c(0,1:7,0),nc=1), height=c(3,table(d$Group),3)) #Plus the height of the empty spaces.
gdist <- as.matrix(gdist)
for (i in 1:N){
par(mar=c(0,3,0,7))
rc_tokeep <- row.names(subset(d, G==i))
if(length(rc_tokeep)>2){
dis <- as.dist(gdist[rc_tokeep, rc_tokeep])
dend <- hclust(dis, method = "average")
plot(as.dendrogram(dend),horiz=TRUE,xlim=c(.8,0),xaxt="n",yaxt="n")
abline(v=seq(0,.8,.1),lty=3)
}else{
plot(NA,xlim=c(.8,0),ylim=c(0,1),axes=F,ann=F)
segments(0,.5,.1,.5)
text(0,.5, pos=4,rc_tokeep,xpd=TRUE)
abline(v=seq(0,.8,.1),lty=3)
}
}
axis(1,at=seq(0,.8,.1))