[R] Expressing a multinomial GLM as a series of binomial GLMs
ccberry at ucsd.edu
Wed Jul 23 06:43:15 CEST 2014
Charles Berry <ccberry <at> ucsd.edu> writes:
> Scherber, Christoph <cscherb1 <at> gwdg.de> writes:
> > Dear all,
> > I am trying to express a multinomial GLM (using nnet) as a series of GLM
> Doing the obvious comparison:
> ggen.preds <-
> sapply( levels(multicats),
> function(x) predict(glm(I(multicats==x)~multiplan1,
> ##  1.349607e-06
> The predictions are the same - up to numerical issues in the algorithms.
But as John Fox points out this does not hold for the general case.
It worked here because the regressor has two categories.
More information about the R-help