How to replace a column in a data frame with another one with a different size

Your
2012/7/8 Michael Weylandt <michael.weylandt at gmail.com>:

On Jul 8, 2012, at 9:31 AM, Stathis Kamperis <ekamperi at gmail.com> wrote:

Hello everyone,

I have a dataframe with 1 column and I'd like to replace that column
with a moving average.
Example:

library('zoo')
mydat <- seq_len(10)
mydat
[1] 1 2 3 4 5 6 7 8 9 10
df <- data.frame("V1" = mydat)
df
  V1
1   1
2   2
3   3
4   4
5   5
6   6
7   7
8   8
9   9
10 10
df[df$V1 <- rollapply(df$V1, 3, mean)]
Error in `$<-.data.frame`(`*tmp*`, "V1", value = c(2, 3, 4, 5, 6, 7, 8,  :
 replacement has 8 rows, data has 10

I'm not sure you need the outer df[...] -- I think you just want

df$V1 <- rollapply(df$V1,3,mean)

However, this will still give you the error message you're seeing because rollapply() only returns 8 values here (you don't get the "endpoints" by default). To get the right number of rows, you want

rollapply(df$V1, 3, mean, fill = NA) # Change NA if desired

which will put NA's on each end and give you a length 10 result, as needed.

Thanks Michael (and arun@)!

If I would do that, then (in my particular case), I'd need to
eliminate NA's, with something like:
df$V1 <- df$V1[!is.na(df$V1)]

which would still fail with the same error message :-P
You're getting tripped up (again) by trying to sub-assign something
that's too small.

df is a rectangular array of data: on the RHS of that expression, you
are selecting out a subset of it of say 8 rows and telling R to
replace the 10-row V1 column with those 8 elements. This cannot be
done with the fixed rectangular structure and hence the error message.

What you want to do is something like this:

df[!is.na(df$V1), ]

Let's walk through that

df$V1 -- take the V1 column of df

is.na() -- get a logical vector saying where NAs are

!is.na() -- identify the rows where there _aren't_ NAs

df[ !is.na(), ] -- (the important one) take the rows of df (all
columns) where there aren't NAs

What you might be wanting to do is

df <- df[!is.na(df$V1), ]

This is much better than what you are trying to do (working on the
whole array at a time and trusting R to keep it all together than
trying to manipulate slices individually)

But even more idiomatic would be

complete.cases(df)

Take a look at some introductory material and try to wrap your head
around indexing rows and columns together again: it's a fantastic
paradigm and will be of much more use to you long run than trying to
work on individual columns for subsetting/data-cleaning.

Best,
Michael
Regards,
Stathis

Best,
Michael

I could use a temporary variable to store the results of rollapply()
and then reconstruct the data frame, but I was wondering if there is a
one-liner that can achieve the same thing.

Best regards,
Stathis

P.S. If you don't mind, cc me at your reply because I'm not subscribed
to the list (but I will check the archive anyway).

______________________________________________
R-help at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.