[R] Autogenerate tags in tag=value pairs for

[Gabor Grothendieck]

Try na.locf from the zoo package and then use merge with specified suffixes:

library(zoo)
f <- function(x) {
   rownames(x) <- NULL
   merge(x, na.locf(x[-1], na.rm = FALSE), by = 0, suffixes = c("", ".by"))[-1]
}
do.call("rbind", by(x, x$id, f))


On 3/7/07, Jon Olav Vik <j.o.vik at bio.uio.no> wrote:
> Dear list,
>
> Is there a way to programmatically specify tag names for the ... (ellipsis)
> part of the argument list to a function? In other words, a way to do this:
>
> x <- data.frame(A=1:5)
>
> if the name "A" was not hardcoded but given by a variable, and without
> resorting to:
>
> x <- data.frame(1:5)
> names(x) <- "A"
>
>
> A longer example describing my actual problem follows. Thanks in advance for
> any help.
>
> Best regards,
> Jon Olav
>
>
> I want to use function transformBy() in package doBy. The key is that the "...
> Further arguments of the form tag=value" require "tag" to be specified,
> otherwise the output does not include the results of my groupwise calculations.
>
> Quoting the documentation:
> " transformBy(doBy)
> " Function to make groupwise transformations of data
> " by applying the transform function to subsets of data.
> "
> " Usage
> " transformBy(formula, data, ...)
> "
> " Arguments
> " formula A formula with only a right hand side, see examples below
> " data A data frame
> " ... Further arguments of the form tag=value
>
> ### example ###
>
> # a function to replace NAs with the last non-NA value from above
> filldown <- function(x) {
>    notna <- !is.na(x) # elements with values
>    ix <- cumsum(notna) # index to previous element (but zeros where we need NA)
>    ix[ix==0] <- NA # use [NA] as index to produce NA in output
>    return(x[notna][ix]) # for each: return previous value if found, else NA
> }
> # illustration of how it works
> tmp <- c(NA,NA,1,NA,3,NA,NA)
> cbind(tmp,filldown(tmp))
>
> # I now want to apply filldown() to subsets of a data frame
> # and I want it to work on several columns
>
> # generate a data frame for illustration,
> # with a few non-NA values scattered round
> set.seed(5) # repeatable example
> x <- data.frame(id = rep(1:4,each=6), v1=NA, v2=NA)
> ix <- which(runif(nrow(x))>0.75)
> x[ix,2] <- rpois(length(ix),5)
> ix <- which(runif(nrow(x))>0.75)
> x[ix,3] <- rpois(length(ix),5)
> x
>
> library(doBy)
> # the hard way -- works as required,
> # but I would like not having to hardcode column names v1 etc.
> transformBy(~id,data=x,v1.fd = filldown(v1),v2.fd = filldown(v2))
>
> # does not work because
> # output includes only columns explicitly mentioned in the ... argument
> transformBy(~id,data=x,function(y) lapply(y,filldown))
>
> ______________________________________________
> R-help at stat.math.ethz.ch 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.
>