I think step-by-step is about the only way (warning: untested code!): x <- as.matrix(xy[, 1:3]) x <- cbind(x, mean.x = apply(x, 1, mean), var.x = apply(x, 1, var)) y <- as.matrix(xy[, 4:6]) y <- cbin
might want to look at http://www.bell-labs.com/project/PORT/ and get the manual. --Original Message-- From: s-news-owner@lists.biostat.wustl.edu [mailto:s-news-owner@lists.biostat.wustl.edu] On Beha
Notes below: --Original Message-- From: s-news-owner@lists.biostat.wustl.edu [mailto:s-news-owner@lists.biostat.wustl.edu] On Behalf Of Leeds, Mark Sent: Friday, 1 April 2005 7:20 AM To: s-news@lists
Don't be too worried about a loop solution if you are not going to be doing it billions of times: for(j in 2:ncol(mat)) mat[ , j] <- mat[ , j] & mat[ , j-1] Warning: untested code. Bill Venables. --O
Here is how I would do it. stirling <- function(n) { v <- c(1, 0) if(n == 0) return(v) for(j in 1:n) v <- c(v, 0) - j*c(0, v) abs(v) } Seems to work OK for quite large n. Bill Venables. --Original M
When there are no missing values I presume the design is then orthogonal, in which case you will have information on each separate effect in one stratum only. (This is the goal of most split-plot des