[R] multiple plots for data sets of inconsistent size

Laura Quinn laura at env.leeds.ac.uk
Fri Jun 18 10:51:51 CEST 2004

I am wanting to plot pca loadings onto sites superimposed on a contour
map. There are a maximum of 20 sites at which loadings might appear -
however due to the nature of my data, missing data has meant that some
stations have not been included in some of the pca.

For example, I am performing pca for 5 different variables at 20 different
sites for a number of different time intervals and want to perform a cross
comparison to see where key trends lie. Where data is not available for
any particular station at any given time interval I am still keen to
analyse the principal components on the others.

I have been using the following code to produce the plots I want:

for (i in 1:4) {

 pc <- p6.5am.pca$rotation[,i]
 pc.pos <- (pc > 0)
 if (any(pc.pos)) {
 if (any(!pc.pos)) {

My problem is this - due to the large number of plots required I would
like to be able to batch process the results, but in the case of applying
the above code when one (or more) station(s) is missing from the pca I get
an error message as the x and y lengths differ. Is there a way I can get the code
to simply return a zero loading at a site if it hasn't been included in
the calculation of the pca? (or is there another better way around this

I hope that this question makes some sense.

Any ideas?

More information about the R-help mailing list