Skip to content

Combining series of variables using identifier

3 messages · Lauri Nikkinen, K. Elo, Henrique Dallazuanna

Original
Lauri Nikkinen
# 
R users,

I have df like this

a <- data.frame(    indx =    1:20,
                    var1 =    rep(c("I20", "I40", "A50", "B60"), each=5),
                    var1_lab= rep(c("cat", "dog", "mouse", "horse"), each=5),
                    var2 =    rep(c("B20", "X40", "D50", "G60"), each=5),
                    var2_lab= rep(c("car", "bicycle", "train", "bus"), each=5))
str(a)

I'd like to create new variables by combining "varX" and "varX_lab" like this:

a$var1_new <- factor(paste(a$var1, a$var1_lab, sep=": "))
a$var2_new <- factor(paste(a$var2, a$var2_lab, sep=": "))
a

But, in the real world, I have multiple df:s and many of variables
named by the same manner. Is there a possibility to create a function
which combines these variables and creates new ones into the same df
using "varX" as an identifier?

Many thanks,
Lauri
K. Elo
# 
Hi,

if the columns always follow the same order (indx,var#, var#_lab ...), 
then You could use the column numbers and do the following:

1) CN<-colnames(df)[#] (#=colnum of 'var#')
2) df$NEW<-(here the expression to create the new variable)
3) colnames(df)[ncol(df)]<-c(paste(CN,"new",sep="_"))

This can easily be built in a loop.

Hope this helps,
Kimmo

Lauri Nikkinen kirjoitti viestiss??n (26.02.2008):
Henrique Dallazuanna
# 
Try this:

foo <- function(data, ...)
{
patt <- grep("var[0-9]$", names(data), value=T)
res <- paste(sapply(data[, patt], as.character), sapply(data[,
paste(patt, "lab", sep="_")], as.character), sep=": ")
cbind(a, matrix(res, ncol=length(patt), dimnames=list(NULL,
paste(patt, "new", sep="_"))))
}

foo(a)
On 26/02/2008, Lauri Nikkinen <lauri.nikkinen at iki.fi> wrote: