Hello all,
I am running a multinomial logistic regression using the mlogit command, but have run into quasi-perfect separation in one of my key predictors. Empirical methods exist for conducting penalized multinomial logistic regressions, but I can't seem to find a command in Stata that supports this. Examples of papers that talk about the penalized multinomial logit include: http://www.utstat.toronto.edu/wordpr...ports/0505.pdf
This has actually been implemented in R, as seen in the pmlr package for that software.
Does anyone know if there is a way to use this method in Stata without having to write a program from scratch?
Thanks,
Carlos
(Not-so-random thought: it would be great, if someone were to write this program, if it could take advantage of the mlogit postestimation commands that already exist.)
I am running a multinomial logistic regression using the mlogit command, but have run into quasi-perfect separation in one of my key predictors. Empirical methods exist for conducting penalized multinomial logistic regressions, but I can't seem to find a command in Stata that supports this. Examples of papers that talk about the penalized multinomial logit include: http://www.utstat.toronto.edu/wordpr...ports/0505.pdf
This has actually been implemented in R, as seen in the pmlr package for that software.
Does anyone know if there is a way to use this method in Stata without having to write a program from scratch?
Thanks,
Carlos
(Not-so-random thought: it would be great, if someone were to write this program, if it could take advantage of the mlogit postestimation commands that already exist.)