Re: [R] separate row averages for different parts of an array

From: Gabor Grothendieck <>
Date: Thu 17 Aug 2006 - 17:53:51 EST

The following reshapes mat so we can take the means of the columns of the resulting 3d array and then transposes it back to the original orientation:

   t(colMeans(array(t(mat), c(100, 448, 24))))

You might want to try it on this test set first where anscombe is an 11x8 data set built into R. Here are 4 solutions using anscombe

  1. This is just the above written for the anscombe data set:

t(colMeans(array(t(anscombe), c(4,2,11))))

2. Here is a solution using apply instead of colMeans and t. In this case anscombe is a data.frame, not an array/matrix, and we need to turn it into one first. The prior solution also required a matrix but tranpose will convert a dataframe to a matrix so we did not have to explicitly do it there. If your array is indeed an array as stated in your post then you can omit the as.matrix part. In your case the c(11,4,2) vector would be c(24, 100, 448) :

apply(array(as.matrix(anscombe), c(11,4,2)), c(1,3), mean)

3. Here is another solution. This one uses the zoo package and does have the advantage of not having to specify a bunch of dimensions. It uses rapply from zoo (which will be renamed rollapply in the next version of zoo so as not to conflict with the new rapply that is appearing in R 2.4.0). In your case both occurrences of 4 would be 100:

coredata(t(rapply(zoo(t(anscombe)), 4, by = 4, mean)))

4. This is Marc's solution except we use seq instead of : at the end in order to make use of the length= argument. In your case c(11, 8, 4) would be c(1, 44800, 100) and length = 4 would be length = 100:

sapply(seq(1, 8, 4), function(i) rowMeans(anscombe[, seq(i, length = 4)]))

On 8/16/06, Spencer Jones <> wrote:
> I have an array with 44800 columns and 24 rows I would like to compute the
> row average for the array 100 columns at a time, so I would like to end up
> with an array of 24 rows x 448 columns. I have tried using apply(dataset, 1,
> function(x) mean(x[])), but I am not sure how to get it to take the average
> 100 columns at a time. Any ideas would be welcomed.
> thanks,
> Spencer
> [[alternative HTML version deleted]]
> ______________________________________________
> mailing list
> PLEASE do read the posting guide
> and provide commented, minimal, self-contained, reproducible code.
> mailing list PLEASE do read the posting guide and provide commented, minimal, self-contained, reproducible code. Received on Thu Aug 17 17:58:16 2006

Archive maintained by Robert King, hosted by the discipline of statistics at the University of Newcastle, Australia.
Archive generated by hypermail 2.1.8, at Thu 17 Aug 2006 - 20:23:14 EST.

Mailing list information is available at Please read the posting guide before posting to the list.