1
$\begingroup$

Is it possible to use the rms package to model multinomials logits, or elsewise to model several binary logits to achieve the same effect? I am aware that there are many other packages specifically designed to fit multinomial logits, but I am quite attached to the suite in rms.

$\endgroup$
  • $\begingroup$ Not the answer you're looking for, but the multinom function in the nnet is superb. Uses the exact same functionality as any glm object (which technically speaking, it is). Aside from preference, what do you need form rms? $\endgroup$ – AdamO Jul 28 '15 at 19:13
  • $\begingroup$ Bootstrapped covariance, prediction plots (with prediction intervals), influence statistics, and splines are some of the advantages I'm after. I'm actually using nnet and I don't have any issues with it—I think it covers my basic needs very well. $\endgroup$ – Johan Larsson Jul 28 '15 at 19:18
0
$\begingroup$

The answer is no. Having delved into this myself, it appears that it is not possible at the moment. However, it could be accomplished in the future were someone smarter than me to put some effort into it.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.