Re: [Rd] Unexpected behavior in factor level ordering

From: Uwe Ligges <ligges_at_statistik.tu-dortmund.de>
Date: Sat, 25 Feb 2012 19:25:49 +0100

On 25.02.2012 19:16, Paul Johnson wrote:
> Hello, Everybody:
>
> This may not be a "bug", but for me it is an unexpected outcome. A
> factor variable's levels
> do not retain their ordering after the levels function is used. I
> supply an example in which
> a factor with values "BC" "AD" (in that order) is unintentionally
> re-alphabetized by the levels
> function.
>
> To me, this is very bad behavior. Would you agree?
>
>
> # Paul Johnson 2012-02-05
>
> x<- c("AD","BC","AD","BC","AD","BC")
> xf<- factor(x, levels=c("BC", "AD"), labels=c("Before Christ","After Christ"))
> y<- rnorm(6)
>
> m1<- lm (y ~ xf )
>
> plot(y ~ xf)
>
> abline (m1)
> ## Just a little problem the line does not "go through" the box
> ## plot in the right spot because contrasts(xf) is 0,1 but
> ## the plot uses xf in 1,2.
>
> xlevels<- levels(xf)
> newdf<- data.frame(xf=xlevels)
>
> ypred<- predict(m1, newdata=newdf)
>
> ##Watch now: the plot comes out "reversed", AC before BC
> plot(ypred ~ newdf$xf)
>
> ## Ah. Now I see:
>
> levels(newdf$xf)
> ## Why doesnt newdf$xf respect the ordering of the levels?

Because xlevels was a character and you coerced it to a factor by calling data.frame(xf=xlevels) on it without telling anything about the orderiung, hence it got sorted lexicographically.

Uwe Ligges

>
>
>



R-devel_at_r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel Received on Sat 25 Feb 2012 - 18:30:32 GMT

This quarter's messages: by month, or sorted: [ by date ] [ by thread ] [ by subject ] [ by author ]

All messages

Archive maintained by Robert King, hosted by the discipline of statistics at the University of Newcastle, Australia.
Archive generated by hypermail 2.2.0, at Sun 26 Feb 2012 - 17:30:21 GMT.

Mailing list information is available at https://stat.ethz.ch/mailman/listinfo/r-devel. Please read the posting guide before posting to the list.

list of date sections of archive