Hello,
If the output of ftable is kept in an object named 'ft', try this.
ftperc <- t(apply(ft, 1, function(x)
if(sum(x)) round(100*x/sum(x)) else rep(0, length(x))))
attributes(ftperc) <- attributes(ft)
ftperc
(And if you need it to process several datasets, a function with 'ft' as
the argument and the code above as its body is straightforward to write.)
Hope this helps,
Rui Barradas
Em 13-05-2012 11:00, r-help-request at r-project.org
escreveu:> Date: Sat, 12 May 2012 14:00:03 -0700 (PDT)
> From: cassiorx<cassiodorus at hotmail.com>
> To:r-help at r-project.org
> Subject: [R] categorical analysis - grouping rows
> Message-ID:<1336856403964-4629503.post at n4.nabble.com>
> Content-Type: text/plain; charset=us-ascii
>
> I apologize up front if this has been covered elsewhere - but I can't
find
> any such question.
>
> I have a data set that contains academic data: term (i.e., semester),
> student id, dept, class, success (1=Y, 0=N)
>
> I want to look at dept by term to determine descriptive statistics for
> success to failure ratios. The intent being to discover if there are
> departments that contribute significantly to the Simpson Paradox, that is,
> that make overall success/failure rates undependable.
>
> It's easy to use ftable to get the counts for what I need (row names
dept
> and success, col name success. So I get something that looks like this:
>
> Term 1st 2nd 3rd 4th 5th
> dept success
> AAA 0 155 240 163 286 293
> 1 424 570 349 582 429
> AAB 0 55 64 103 46 109
> 1 122 117 145 112 145
> AAC 0 11 3 4 4 4
> 1 19 12 23 11 7
>
> How can I calculate percentages by dept so that I get
>
> AAA 0 27 ....
> 1 73 ....
> AAB 0 ...
>
> Part of my lack of understanding is that I don't see a way to get the
dept
> (by term) totals into a data structure that I can use to calculate the
> percentages. I can write procedural code to do this but is there some r-way
> that would be better?
>
> --
> View this message in
context:http://r.789695.n4.nabble.com/categorical-analysis-grouping-rows-tp4629503.html
> Sent from the R help mailing list archive at Nabble.com.
>