Skip to content

how to get colnames of a dataframe within a function called by 'apply'

3 messages · Leo Gürtler, Gabor Grothendieck

Original
Leo Gürtler
# 
Hello alltogether,

how is it possible to assign the colnames of a data.frame to a function 
called by apply, e.g. for labeling a plot?
Example: I want to plot several qqnorm-plots side by side and there 
should be a maintitle for each qqnorm-plot which is identical to the 
respective colname.
I checked, but the column which is processed by the function called by 
apply does not contain a colname (because by using str() it seems it is 
no column at this point, but just a e.g. numerical vector).
I also tried with colnames() from within the function, but was not 
successful to get the apropriate colname - either the whole string or 
just the first one.Thus it lacks of a counter that contains the number 
of the row which is processed.

Here is an example code:

<---snip--->

nv <- function(xno.na)
{
  par.or <- par(mfrow=c(3,3))
  qqnorm(xno.na, main="HERE SHOULD BE THE NAME OF THE COLUMN")
  qqline(xno.na, col="red")
  par(par.or)
  print(str(xno.na))
}

 > temp # just a part of the whole data.frame
        klarb1    klarb2 abarb laut skla1 skla2
a      NA 13.068182   7.5    4   0.5   0.5
b      NA  6.818182   9.0    6    NA   0.5
c      15.11628  6.818182  10.0    4   1.0   1.5
d      NA 18.181818  19.0    2   1.0   0.5

 > apply(temp,2,nv)

</---snip--->


Thanks a lot!

best wishes,

leo
Gabor Grothendieck
# 
Iterate over the column names rather than the columns themselves:

par.or <- par(mfrow=c(3,3))
nv <- function(name, data) {
	qqnorm(data[,name], main=name)
	qqline(data[,name], col="red")
	invisible()
}
sapply(colnames(temp), nv, data = temp)
par(par.or)
On 10/28/05, Leo G??rtler <leog at anicca-vijja.de> wrote:
Gabor Grothendieck
# 
One other comment.  A for loop would seem to be
appropriate here since you are not using the result
of sapply anyways:

par.or <- par(mfrow = c(3,3))

for(n in colnames(temp)) {
  qqnorm(temp[,n], main = n)
  qqline(temp[,n], col = "red")
}

par(par.or)
On 10/28/05, Gabor Grothendieck <ggrothendieck at gmail.com> wrote: