Reply With Quote
I don't think it can be done. Multinomial logistic regression with k levels of the categorical response variable works by comparing (k-1) of the levels to the one remaining level. I don't think it works any other way.
Does anyone knows how I can formulate a the link function of a multinomial logistic regression without using a base class
teotjunk
I don't think it can be done. Multinomial logistic regression with k levels of the categorical response variable works by comparing (k-1) of the levels to the one remaining level. I don't think it works any other way.
yes i agree with squareandre
Posting Permissions
|
|
