I'm applying multinomial logistic regression to examine the influence of various predictors on three outcomes. The issue I'm facing now is that the SPSS outputs returned extremely large positive/negative values of some predictor values, that is, "quasi-complete seperation". I want to deal with this problem using penalized logistic regression since I do not want to remove the variables. I would greatly appreciate it if you could give me advice on how to perform penalized logistic regression or recommend statistics services. Thank you.
Hello Atikhom Thienthong Since you mentioned the name of the SPSS software, I write that, An approach due to David Firth has been implemented in an R package named logistf. there is an SPSS Statistics R-based extension named stats firthlog that allows SPSS Statistics users to access this package.
I have given it in the attached picture.
Abolfazl Ghoodjani Thank you for your help. It seems like it is beyond my knowledge. I greatly appreciate it if you could give me a step-by-step advice on how to install the extension and run the data. Thank you.
- Badges
- Science topic