ripley at stats.ox.ac.uk
2007-Mar-29 10:59 UTC
[Rd] (PR#9578) rbind.data.frame reacts on levels without
In your data frames, 'y' is a category and the following comment in the code makes clear this is deliberate. if( !is.null(levels(xj)) ) { all.levs[[j]] <- levels(xj) facCol[j] <- TRUE # turn categories into factors } else facCol[j] <- is.factor(xj) The behaviour is compatible with S apart from the comment in the R help page. That categories are promoted to factors is consistent with the help for 'category'. It is not a bug, but the help page could say what happens for categories. On Fri, 23 Mar 2007, Lutz Prechelt wrote:> I am sorry. > Here is a minimal example: > > a=data.frame(x=c(1:3),y=unclass(factor(c(5:7)))) > b=data.frame(x=c(11:13),y=unclass(factor(c(7:9)))) > rbind(a,b) > > Lutz Prechelt > >-- Brian D. Ripley, ripley at stats.ox.ac.uk Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/ University of Oxford, Tel: +44 1865 272861 (self) 1 South Parks Road, +44 1865 272866 (PA) Oxford OX1 3TG, UK Fax: +44 1865 272595