[Rd] Bug in cor function (PR#7689)

[Peter Dalgaard]

cig69410 at syd.odn.ne.jp writes:

> I can't hardly accept the result of  cor function with 
> pairwize.colplete.obs or complete.obs
> 
> insert print statements in cor function,
> 
> 
> +     if (method != "pearson") {
>   +         Rank <- function(u) if (is.matrix(u))
>   +             apply(u, 2, rank, na.last = "keep")
>   +         else rank(u, na.last = "keep")
>   +         x <- Rank(x)
>   +         print(x) # add
>   +         if (!is.null(y)) {
>   +             y <- Rank(y)
>   +             print(y) # add
>   +         }
>   +     }
>   +     .Internal(cor(x, y, na.method, method == "kendall"))
> 
> and, data is
>   > x <- c(7, 9, 8,  0, NA, NA)
>   > y <- c(2, 3, 4, NA,  4,  3)
> 
> and, call cor function
>   > cor(x, y, use="pair", method="sp")
> 
> order of x, and y are
>   [1]  2  4  3  1 NA NA
>   [1] 1.0 2.5 4.5  NA 4.5 2.5
> 
> alas!! and the result is
>   [1] 0.4271211
> 
> oh! no!!
> 
> the result must be 0.5

And which part of the following did you fail to understand?

      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")'. Notice also that the ranking is (currently) done
     removing only cases that are missing on the variable itself,  which
     may not be what you expect if you let 'use' be '"complete.obs"' or
     '"pairwise.complete.obs"'.

If you have improved code to contribute, you're welcome (notice that
this requires reranking for every pair of variables in an n x n
correlation matrix in the pairwise case), but there's really not much
point in reporting issues that are already known and documented.

-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)             FAX: (+45) 35327907