CODEWITHSUNDEEP
rhierarchical-clusteringcoercion
Hanjo Odendaal
Hanjo Odendaal

Reputation: 1441

Coerce HCPC object to hclust to use library(ape)

I am currently performing hierarchical cluster. I opted to use FactoMineR due to its nice data exploration pipeline: data->MFA->HCPC.

data("wine")
names(wine)
res<-MFA(wine,group = c(2,5,3,10,9,2),type=c("n",rep("s",5)),
         ncp=5,name.group = c("orig","olf","vis","olfag","gust","ens"),
         num.group.sup = c(1,6))

res.hcpc<-HCPC(res, nb.clust=0, consol=F, iter.max=10, min=3, 
           max=NULL, metric="euclidean", method="ward", order=TRUE,
           graph.scale="sqrt-inertia", nb.par=5, graph=TRUE, proba=0.05, 
           cluster.CA="rows",kk=Inf)

#Now for some nice plots
plot(res.hcpc,draw.tree = T,choice = "tree")
plot(res.hcpc,draw.tree = T,choice = "bar")

These are more than sufficient when wanting to extract answers, but I am looking to present these trees. i.e I want them a bit spruced up through library(ape) - examples

The problem is that these functions require a hclust object. One option I had was to use the MFA results and use an hclust function as: res.hc<-hclust(dist(res$global.pca$ind$coord),method = "ward.D2") to build a hclust object.

This does work, however I find the information that HCPC gives is much more informative (and answers differ from hclust). Thus I want to work with the HCPC object and not hclust, but still be able to use the functions in library(ape). Any idea how one would coerce HCPC to hclust?

Upvotes: 2

Views: 525

Answers (1)

chasemc
chasemc

Reputation: 869

for library(ape)

hcpcTree<-res.hcpc$call$t$tree
apeTree<-as.phylo(hcpcTree)

for library(stats)

dendextendTree<-as.phylo.dendrogram(hcpcTree)

Upvotes: 2

Related Questions