Rank correlation matrix in R

Question

How do I produce a rank correlation matrix in an elegant way in R given a data frame with many columns? I couldn't find a built-in function, so I tried

> test=data.frame(x=c(1,2,3,4,5), y=c(5,4,3,2,1))
> cor(rank(test))

(only 2 columns for simplicity, real data has 5 columns) which gave

> Error in cor(rank(test)) : supply both 'x' and 'y' or a matrix-like 'x'

I figured that this was because rank takes a single vector. So then I tried

> cor(lapply(test,rank))

to get rank applied to each column in the data frame, treating the data frame as a list, which gave the error

> supply both 'x' and 'y' or a matrix-like 'x'

and I finally ended up getting something working with

> cor(data.frame(lapply(test,rank)))
   x  y
x  1 -1
y -1  1

However this seems pretty verbose and ugly. I'm thinking there must be a better way -- if so what?

Answer 1

You are doing it wrong -- use the kendall method argument for cor() instead:

R> testdf <- data.frame(x=c(1,2,3,4,5), y=c(5,4,3,2,1))  
R> cor(testdf, method="kendall") 
   x  y 
x  1 -1    
y -1  1   
R> 

From help(cor):

For cor(), if method is "kendall" or "spearman", Kendall's tau or Spearman's rho statistic is used to estimate a rank-based measure of association. These are more robust and have been recommended if the data do not necessarily come from a bivariate normal distribution. For cov(), a non-Pearson method is unusual but available for the sake of completeness. Note that "spearman" basically computes cor(R(x), R(y)) (or cov(.,.)) where R(u) := rank(u, na.last="keep"). In the case of missing values, the ranks are calculated depending on the value of use, either based on complete observations, or based on pairwise completeness with reranking for each pair.