What kind of prior is used in mnrfit?

what is the prior used by the mnrfit algorithm?

Answer 1

[the cyclist]

Maybe I am displaying massive ignorance here, but multinomial logistic regression (which is what mnrfit does) is not a Bayesian technique, and requires no prior. Right?

[Peter Perkins]

cyclist, you are not displaying ignorance. mnrfit uses maximum likelihood, so no prior. It's essentially a generalization of a binomial logistic regression as done by glmfit.

[Mr M.]

OK, this means uniform prior? but the possible range is infinite for the parameters

[the cyclist]

No. It means the method used to calculate the results does not require any assumptions about a prior probability distribution. There is no prior.