[R] plotting different level of classficifaction

[Jim Lemon]

jiho.han wrote:
> hello, useRs~
> suppose i have a matrix as follows:
> item    category      sub-category
>   A           1                  11
>   B           1                  12
>   C           1                  12
>   D           2                  21
>   E           2                  22
> 
> i like to draw a plot that represent the proportion of each 
> category at each level. that is, i want a plot like this:
>     +-----+   +-----+  
>     |        |   |   11 |
>     |        |   +-----+
>     |   1   |   |   12  |
>     |        |   |        |
>     +-----+   +-----+
>     +-----+   +-----+
>     |        |   |   21 |
>     |   2   |   +-----+
>     |        |   |   22 |
>     +-----+   +-----+
>   category   sub-cateory
> 
> hope my plot makes sense. anyone know how to make such plot?
> any hint/suggestion would be much appreciate. thank you.
> 
Hi jiho,
An interesting request. Here is a very basic function that does what you 
want (I think) with a slightly expanded dataset to show its generality.

item cat subcat subsubcat
A 1 11 111
B 1 12 121
C 1 12 122
D 2 21 211
E 2 22 221
F 3 31 311
G 3 31 312
H 3 32 322
I 3 32 322
J 3 33 331
K 3 34 341
L 3 34 341
M 3 34 342

bh<-read.table("bh.dat",header=TRUE)

barhier<-function(x) {
  dimx<-dim(x)
  plot(0,xlim=c(0,dimx[2]),ylim=c(0,dimx[1]),axes=FALSE,type="n",
   xlab="",ylab="")
  for(level in 1:dimx[2]) {
   top<-dimx[1]
   levelfreq<-freq(x[,level])
   valuenames<-names(levelfreq[[1]])
   for(value in 1:length(levelfreq[[1]])) {
    bottom<-top-levelfreq[[1]][value]
    rect(level-1,bottom,level,top)
    text(level-0.5,(bottom+top)/2,valuenames[value])
    top<-bottom
   }
  }
}

barhier(bh[,-1])

Jim