Hierarchical Clustering 

In[1]:=

X Needs["RLink`"] InstallR[]

Start with some matrix data.

In[2]:=

X mdata = RandomVariate[NormalDistribution[], {10, 5}]

Out[2]=

Send the data to R.

In[3]:=

X RSet["y", mdata]

Out[3]=

Add labels to columns and rows.

In[4]:=

X REvaluate[" { dimnames(y)<- list(paste(\"g\", 1:10, sep=\"\"), paste(\"t\", 1:5, \ sep=\"\")) y }"]

Out[4]=

Construct correlation and distance matrices.

In[5]:=

X REvaluate[" { c <- cor(t(y), method=\"spearman\") d <- as.dist(1-c) }"]

Out[5]=

Call the hierarchical clustering routine.

In[6]:=

X REvaluate["hr <- hclust(d, method = \"complete\", members=NULL)"]

Out[6]=

Define plotting functions from the example on higher-order functions.

In[7]:=

X mathematicaRPlotWrapper = RFunction["function(filename, plotfun){ pdf(filename) plotfun() dev.off() }"]

Out[7]=

In[8]:=

X getRPlot[plotFun_RFunction] := With[{tempfile = FileNameJoin[{$TemporaryDirectory, "temp.pdf"}]}, If[FileExistsQ[tempfile], DeleteFile[tempfile]]; mathematicaRPlotWrapper[tempfile, plotFun]; If[! FileExistsQ[tempfile], Return[$Failed]]; Import[tempfile] ];

Produce the dendrogram plots (the "hang" parameter controls the way the dendrogram is plotted).

In[9]:=

X Show[#, ImageSize -> Large, PlotRange -> All] &@ getRPlot@RFunction["function(){ par(mfrow = c(2, 2)) plot(hr, hang = 0.1) plot(hr, hang = -0.1) }"]

Out[9]=