On 03/06/2013 12:45 AM, km wrote:> Dear All,
>
> I have a table as following
> position type count
> 1 2 100
> 1 3 51
> 1 5 64
> 1 8 81
> 1 6 32
> 2 2 41
> 2 3 85
> and so on
>
>
> Normally if would have a vector of 2,3,4,5... by position position and
> plot them by position.
> But now i have counts of these types.
> Is there a way to compute boxplot of such kind of data ?
Hi KM,
We must assume that the "type" variable is to be used as a value,
otherwise you would want something like a frequency plot by two factors
of position and type. (If this is the case, I would suggest a nested bar
plot). Here is a fairly awful kludge that will get you a boxplot (tdf is
your "table"):
reprow<-function(x) return(matrix(rep(x[1:2],x[3]),ncol=2,byrow=TRUE))
replist<-apply(as.matrix(tdf),1,reprow)
repmat<-replist[[1]]
for(rep in 1:length(replist)) repmat<-rbind(repmat,replist[[rep]])
boxplot(repmat[,2],repmat[,1])
With apologies
Jim