[R] Econometrics ...
ben at zoo.ufl.edu
Fri Oct 12 20:59:57 CEST 2001
My two cents --
Constructing and maintaining an elaborate alias structure just so people
can write logit(y ~ x) rather than glm(y~x,link=logit) [etc.] seems like a
bad idea to me. A "meta-package" (as Greg Warnes suggested), and a small
discipline-specific document like those Jonathan Baron and Pete St. Onge
have written or started to write for psychology and ecology (see the
Contributed Documentation on CRAN), seem like a better way to go. (The
package could also include useful discipline-specific tools that don't
already exist.) Among other problems, we're all going to go crazy when
people start to write in to the R help list for help with the logit(),
ols(), etc., etc., functions that none of us have heard of. [It also
might not be an entirely bad thing if econometricians (for example)
learned that the rest of the world has something called a GLM that does
what they know of as logit and probit models.]
On Fri, 12 Oct 2001, Ott Toomet wrote:
> I think an econometric package is needed. R is more oriented to general
> statistics (at least from my viewpoint) and lacks some commonly used
> econometric methods, some other methods are spread in different packages.
> Third problem is that many functions use algorithms, not common in econometrics
> (e.g. neural networks).
> I think we need at first a package wich is able to do a most basic
> econometrics (I am not a time series man, so I drop the time series stuff,
> these are just problems I have encountered):
> econometrical modelling including
> ols (alias to lm())
> logit, probit (alias to glm()?)
> nlls (glm()?)
> perhaps multinominal logit, using ML method
> tobit model
> and the common tests including white heteroscedastisity, White
> Heteroscedasticity-consistent t-values, Durbin-Watson etc..
> These functions and tests should be in a form, common and easily
> understandable for economists and students, eg to write
> model <- logit( y~x)
> summary( model, white.t=TRUE)
> I think the functions, given in other packages should somehow repeated, at
> least when the method is a bit unfamiliar for econometrics. Of course, this
> package cannot include everything in econometrics, but perhaps enough for a
> basic course. Additionally, it can made ,,covers'' of e.g. tseries package,
> aliasing the functions there in a more common form, but I think it is not a
> main issue.
> I could well help with the functions I have, but unfortunately I cannot
> maintain a whole new package right now.
> best wishes,
> Ott Toomet
> > John Janmaat wrote:
> > >
> > > Hello,
> > >
> > > Is there a package available for R which generates output commonly used
> > > by econometricians (eg., the Durbin-Watson statistic for serial
> > > correlation in regression residuals)? I'm pretty sure most of the stuff
> > > is out there in assorted packages, under different names. However, it
> > > would make my life, and that of my students, easier if it was all in one
> > > package.
> > >
> > > Thanks,
> > >
> > > John.
> > > -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
> > > r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
> > > Send "info", "help", or "[un]subscribe"
> > > (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch
> > > _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
> r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
> Send "info", "help", or "[un]subscribe"
> (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch
318 Carr Hall bolker at zoo.ufl.edu
Zoology Department, University of Florida http://www.zoo.ufl.edu/bolker
Box 118525 (ph) 352-392-5697
Gainesville, FL 32611-8525 (fax) 352-392-3704
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch
More information about the R-help