Skip to content

min, median, mean on data frame (matrix)

2 messages · ivo welch, Gabor Grothendieck

#
dear r experts:

I know its almost surely documented.  if d is a data frame

median(d)  <-- fails (message is "need numeric data")
min(d)  <-- succeeds, but gives a scalar.
mean(d)  <-- gives a vector

I am wondering whether this could be changed into something more
consistent.  most naturally, the functions might just always return a
vector.

of course, if this change were made and if I wanted the old min() and
max() behavior on a data frame, I would need to first flatten the
data.

just a thought...

/iaw
#
One can use sapply to get a consistent set of idioms despite the underlying
inconsistency of the functions themselves (using the builtin dataset anscombe):

sapply(anscombe, median)
sapply(anscombe, min)
sapply(anscombe, mean)
On 12/28/06, ivo welch <ivowel at gmail.com> wrote: