[Rd] Changes in the MEMSS package (data sets from Pinheiro and Bates, 2000, without the groupedData classes)
ggrothendieck at gmail.com
Tue Jul 24 18:22:55 CEST 2007
On a related note CO2 and ChickWeight in datasets have nlme
specific attributes so either those datasets themselves should be
moved to nlme or the nlme specific attributes removed in datasets
On 7/24/07, Douglas Bates <bates at stat.wisc.edu> wrote:
> On 7/24/07, Gabor Grothendieck <ggrothendieck at gmail.com> wrote:
> > Does this include datasets such as CO2 and ChickWeight
> > which are in the datasets package?
> > Could you post a list of the specific datasets you are referring to
> > so there is no confusion what this is about.
> I am only referring to the datasets in the MEMSS package. There will
> be CO2, ChickWeight, Theoph, etc. datasets in the MEMSS package
> without the groupedData classes. However, the versions in the
> datasets package will stay as they are unless and until R-core decides
> to change them.
> Having datasets in the datasets package and in the MEMSS package with
> the same name will produce a warning when you attach the MEMSS
> package. However, I generally use datasets with a call like
> data(Rail, package = "MEMSS")
> which doesn't cause a warning (it is assumed that if you are this
> explicit about the data set then you probably want exactly that
> version and haven't accidently overridden the other).
> The bottom line is that unless you have been in the habit of attaching
> the MEMSS package or getting data sets from it as above then you won't
> notice any difference.
> > On 7/24/07, Douglas Bates <bates at stat.wisc.edu> wrote:
> > > Some time ago Deepayan and I created a package called MEMSS for the
> > > data sets from the nlme package as data frames but not groupedData
> > > objects. Because of advances that Deepayan has made in lattice
> > > graphics many of the specialized plotting methods for the groupedData
> > > objects are no longer needed. It is easier and less confusing to
> > > store these data sets as data frames rather than as groupedData
> > > objects.
> > >
> > > I propose making two further changes in these objects. Those objects
> > > typically had factors converted to ordered factors in an artificial
> > > way and I plan to revert them to factors. Also, I will relabel any
> > > factors with (26 or fewer) numeric levels with letters and to remove
> > > the "ginfo" attribute.
> > >
> > > Please tell me if this will cause you hardship because you have used
> > > these data sets in your work and depend on the current formulation.
> > >
> > > ______________________________________________
> > > R-devel at r-project.org mailing list
> > > https://stat.ethz.ch/mailman/listinfo/r-devel
> > >
More information about the R-devel