Efficient calculation of matrix cumulative standard deviation in r
- by Abiel
I recently posted this question on the r-help mailing list but got no answers, so I thought I would post it here as well and see if there were any suggestions.
I am trying to calculate the cumulative standard deviation of a matrix. I want a function that accepts a matrix and returns a matrix of the same size where output cell (i,j) is set to the standard deviation of input column j between rows 1 and i. NAs should be ignored, unless cell (i,j) of the input matrix itself is NA, in which case cell (i,j) of the output matrix should also be NA.
I could not find a built-in function, so I implemented the following code. Unfortunately, this uses a loop that ends up being somewhat slow for large matrices. Is there a faster built-in function or can someone suggest a better approach?
cumsd <- function(mat)
{
retval <- mat*NA
for (i in 2:nrow(mat)) retval[i,] <- sd(mat[1:i,], na.rm=T)
retval[is.na(mat)] <- NA
retval
}
Thanks.
