Open athowes opened 3 years ago
Could look at including some fixed / random effects only in certain categories. Correspondence also discussed in On the "Poisson Trick" and its Extensions for Fitting Multinomial Regression Models. Structures include: baseline category, nested, etc.
http://www.statsathome.com/2017/09/14/visualizing-the-multinomial-in-the-simplex/ http://www.statsathome.com/2017/08/10/building-the-ilr-from-the-alr-transform/ http://www.statsathome.com/2017/08/09/we-can-do-better-than-the-alr-or-softmax-transform/
Could look at including some fixed / random effects only in certain categories. Correspondence also discussed in On the "Poisson Trick" and its Extensions for Fitting Multinomial Regression Models. Structures include: baseline category, nested, etc.