Skip to content

mean-aggregate – but use unique for factor variables

2 messages · Martin Batholdy, Bert Gunter

Martin Batholdy
# 
Hi,

I have a data.frame which I want to aggregate.
There are some grouping variables and some continuous variables for which I would like to have the mean.

However there are also some factor-variables in the data-frame that are not grouping variables and I actually would like to aggregate these variables with the unique() function.

Is that possible with the standard aggregate-function?

If I use it like this:

aggregate(data, by=list(data$grouping1, data$grouping2), FUN=mean)

I get vectors of NAs for the other factor-variables in the data.frame.
But I actually want the unique value ?

Is that possible?


thanks!
Bert Gunter
# 
In the absence of data via dput() or other wise ....

I would assume that if you made your function FUN something like:

function(x) if(is.factor(x)) unique(x) else mean(x)

(or maybe length(unique(x)) as first clause)

you'd get what you want.  No?

On Tue, Sep 25, 2012 at 2:32 PM, Martin Batholdy
<batholdy at googlemail.com> wrote: