[R] discrete ECDF

[Charles C. Berry]

On Wed, 4 Aug 2010, David Winsemius wrote:

> Dear list;
>
> I just created a utility function that replicates what I have done in the 
> past with Excel or OO.org by putting a formula of the form =sum($A1:A$1) in 
> an upper-corner of a section and then doing a "fill" procedure by dragging 
> the lower-rt corner down and to the right. When divided by the grand sum of 
> the entries this function then calculates a 2D-discrete-ECDF.
>
> I keep thinking I am missing the obvious, but I did try searching. Here is my 
> effort at creating that functionality:
>
> ecdf.tbl <- function (.dat) { .dat <- data.matrix(.dat)  #speeds up 
> calculations
>       .sdat <- matrix(0, nrow(.dat), ncol(.dat) )
>       .sdat[] <- sapply(1:ncol(.dat), function(x)
>                      sapply(1:nrow(.dat),
>                            function(y)  sum(.dat[1:y, 1:x], na.rm=TRUE )  ) 
> )
> return(.sdat) }
>

ecdf.tbl2 <-
 	function(mat) {
 		mat[is.na(mat)] <- 0
 		t( apply( apply( mat,2, cumsum ), 1, cumsum ))}

HTH,

Chuck


>>  tst <- read.table(textConnection("NA 5 6
> 4	5	7
> 5	6	8
> 6	7	9
> NA 8 NA")   )
>
>>  tst
> V1 V2 V3
> 1 NA  5  6
> 2  4  5  7
> 3  5  6  8
> 4  6  7  9
> 5 NA  8 NA
>
>>  ecdf.tbl(tst)
>    [,1] [,2] [,3]
> [1,]    0    5   11
> [2,]    4   14   27
> [3,]    9   25   46
> [4,]   15   38   68
> [5,]   15   46   76
>
>>  ecdf.tbl(tst)/sum(tst, na.rm=TRUE)
>          [,1]       [,2]      [,3]
> [1,] 0.00000000 0.06578947 0.1447368
> [2,] 0.05263158 0.18421053 0.3552632
> [3,] 0.11842105 0.32894737 0.6052632
> [4,] 0.19736842 0.50000000 0.8947368
> [5,] 0.19736842 0.60526316 1.0000000
>
>
> Did I miss a more compact vectorized or sweep()-ed solution? (I realize this 
> is not really a function in the sense that ecdf() is.) I have seen prop.table 
> and margin.table, but could not see how they would address this problem.
>
> -- 
>
> David Winsemius, MD
> West Hartford, CT
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>

Charles C. Berry                            (858) 534-2098
                                             Dept of Family/Preventive Medicine
E mailto:cberry at tajo.ucsd.edu	            UC San Diego
http://famprevmed.ucsd.edu/faculty/cberry/  La Jolla, San Diego 92093-0901