Multinom. The first formula also specifies the response variable. The variables on the rhs of the formula should be roughly scaled to 0 1 or the fit will be slow or may not converge at all.
The variables on the rhs of the formula should be roughly scaled to 0 1 or the fit will be slow or may not converge at all. A log linear model is fitted with coefficients zero for the first class. The first formula also specifies the response variable.
This page uses the following packages.
Multinomial regression is an extension of binomial logistic regression. Also minus twice log likelihood. Make sure that you can load them before trying to run the examples on this page. The residual deviance compared to the full saturated model that explains individual observations exactly.