How to loop through and modify multiple data frames in R

Question

I have data frames A, B, C, ... and want to modify each data frame in the same way, e.g. re-ordering factors levels of a factor which is present in all of the data frames:

A = data.frame( x=c('x','x','y','y','z','z') )
B = data.frame( x=c('x','y','z') )
C = data.frame( x=c('x','x','x','y','y','y','z','z','z') )

A$x = factor( A$x, levels=c('z','y','x') )
B$x = factor( B$x, levels=c('z','y','x') )
C$x = factor( C$x, levels=c('z','y','x') )

This gets laborious if there are lots of data frames and/or lots of modifications to be done. How can I do it concisely, using a loop or something better? A straightforward approach like

for ( D in list( A, B, C ) ) {
D$x = factor( D$x, levels=c('z','y','x') )
}

does not work, because it doesn't modify the original data frames.

EDIT: added definitions of A, B, and C to make it reproducible.

Could you provide [reproducible example](http://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example)? — zero323, Nov 02 '13 at 03:14
Definitions of A, B, and C have been added so that you can run the code. — baixiwei, Nov 02 '13 at 03:26
Thanks. I know it can annoying especially when situation is obvious but it is a good practice and makes our lives easier :) — zero323, Nov 02 '13 at 06:31

score 4 · Answer 1 · answered Nov 02 '13 at 03:27

4

One thing to note about R is that, with respect to assignment, <- is transitive, whereas = is not. Thus, if your data frames are all the same in this respect, you should be able to do something like this:

A$x <- B$x <- C$x <- factor( C$x, levels=c('z','y','x') )

answered Nov 02 '13 at 03:27

gung - Reinstate Monica

11,583
7
60
79

Thanks for the reply. Unfortunately it will not work for my case (I think) because the data frames all have different numbers of rows. I'll modify my example to make this clear. – baixiwei Nov 02 '13 at 03:30
1

This is why the reproducible example is necessary. – gung - Reinstate Monica Nov 02 '13 at 03:32

zero323 · Answer 2 · 2013-11-02T14:59:39.030

3

If you don't need explicit loop you can use lapply:

ll <- lapply(
    list(A, B, C),
    function(df) {
        df$x <- factor(df$x, levels=c('z', 'y', 'x'))
        return(df)
    }
)

Since data is only copied you'll have to use list returned from lapply.

Edit

dfs <- list('A', 'B', 'C')
levels <- c('z', 'y', 'x')

l <- lapply(
    dfs,
    function(df) {
        # Get data frame by name
        df <- get(df)
        df$x <- factor(df$x, levels=levels)
        return(df)
    }
)


for ( i in 1:length(dfs)) {
    assign(dfs[[i]], l[[i]])
}

edited Nov 02 '13 at 14:59

answered Nov 02 '13 at 06:29

zero323

322,348
103
959
935

If you don't put in `return(df)` you will not get back dataframe elements. – IRTFM Nov 02 '13 at 07:18
This is OK, but I would like a way to modify the original data frames, or more precisely, I want to continue referring to them by their original names. Is there an easy way to get that result using the output of this solution? – baixiwei Nov 02 '13 at 14:03
I've posted an edit with example solution but I cannot say I like it. – zero323 Nov 02 '13 at 14:56

How to loop through and modify multiple data frames in R

2 Answers2

Linked