R help - Apr 2004 - rbind with missing columns

Hello,

I have several data sets, which I want to combine column-by-column using
"rbind" (the data sets have ~100 columns). The problem is, that in
some
data sets some of the columns are missing.

Simply using rbind gives me:
"Error in match.names(clabs, names(xi)) : names don't match previous
names"

Is there a simple way to make R do the rbind despite the missing columns
and to fill out the missing data with NA's? (I could program this
somehow, but probably there is one very simple solution available)

To make it clear here a simplified example. "unknown.command" is what
I
am looking for.

A <- data.frame(a = c(1,2,3), b = c(4,5,6))
B <- data.frame(a = c(1,2,3))
unknown.command(A, B) - should give

A B
1 4
2 5
3 6
4 NA
5 NA
6 NA

Thank you for your help
Klaus

On Fri, 30 Apr 2004 13:47:35 +0200, <klaus.thul at infineon.com> wrote :
>Hello,
>
>I have several data sets, which I want to combine column-by-column using
>"rbind" (the data sets have ~100 columns). The problem is, that in
some
>data sets some of the columns are missing.
>
>Simply using rbind gives me:
>"Error in match.names(clabs, names(xi)) : names don't match
previous
>names"
>
>Is there a simple way to make R do the rbind despite the missing columns
>and to fill out the missing data with NA's? (I could program this
>somehow, but probably there is one very simple solution available)
It's not there already as far as I know, but it's not too hard to
write a new rbind that does this:

newrbind <- function(A,B) {
  anames <- names(A)
  bnames <- names(B)
  notinb <- anames[!(anames %in% bnames)]
  if (length(notinb)) B[[notinb]] <- rep(NA, nrow(B))
  notina <- bnames[!(bnames %in% anames)]
  if (length(notina)) A[[notina]] <- rep(NA, nrow(A))
  rbind(A, B)
}

This only works on data.frames; if you want it to work as generally as
the real rbind, you'll need to add some extra conversions at the
start.

Duncan Murdoch

At 11:03 01/05/04, you wrote:
>Message: 3
>Date: Fri, 30 Apr 2004 13:47:35 +0200
>From: <klaus.thul at infineon.com>
>Subject: [R] rbind with missing columns
>To: <r-help at stat.math.ethz.ch>
>Message-ID:
>         <7509DD89A305F34E9EF16F1EEDFB80AF08D6D3 at
drsse401.eu.infineon.com>
>Content-Type: text/plain;       charset="us-ascii"
>
>Hello,
>
>I have several data sets, which I want to combine column-by-column using
>"rbind" (the data sets have ~100 columns). The problem is, that in
some
>data sets some of the columns are missing.
>
>Simply using rbind gives me:
>"Error in match.names(clabs, names(xi)) : names don't match
previous
>names"
>
>Is there a simple way to make R do the rbind despite the missing columns
>and to fill out the missing data with NA's? (I could program this
>somehow, but probably there is one very simple solution available)
>
>To make it clear here a simplified example. "unknown.command" is
what I
>am looking for.
>
>A <- data.frame(a = c(1,2,3), b = c(4,5,6))
>B <- data.frame(a = c(1,2,3))
>unknown.command(A, B) - should give
>
>A B
>1 4
>2 5
>3 6
>4 NA
>5 NA
>6 NA
Does

A$id <- 1:nrow(A)
B$id <- 1:nrow(B) + nrow(A)
AandB <- merge(A, B, all = TRUE)


 > A
   a b id
1 1 4  1
2 2 5  2
3 3 6  3
 > B
   a id
1 1  4
2 2  5
3 3  6
 > AandB
   a id  b
1 1  1 4
2 1  4 NA
3 2  2 5
4 2  5 NA
5 3  3 6
6 3  6 NA
 >

Give you what you wanted? You can always strip out the "id" column if
you
wish (after sorting on it if you would like the original order back.
>Thank you for your help
>Klaus
Michael Dewey
m.dewey at iop.kcl.ac.uk