topic Producing predicted probabilities with a mixed multinomial logit model in Statistical Procedures

Producing predicted probabilities with a mixed multinomial logit model

mattteale — Fri, 26 Aug 2016 12:41:32 GMT

Hi,

I am going through an example where I want to model a situation where people are asked to choose between travaling by auto, plane or public transit. My covariates contain data on the individual like age as well as alternative specific variables like time taken to travel by auto, plane and public transit. I want to run a mixed model that incoporates both types of covariates. If there are j invidiauls in my dataset is it possible to fed my dataset back into the model and generate the probability of the jth person choosing between auto, plane and transit modes of transport?

Thanks in advance

Re: Producing predicted probabilities with a mixed multinomial logit model

Rick_SAS — Fri, 26 Aug 2016 13:28:32 GMT

If I understand your question correctly, I think you can use the OUTPUT statement in PROC GLIMMIX and the PREDICTED keyword to get what you want. See this recent related post about visualizing fitted mixed models. Assuming that you want to model the response as a multinomial proportion, you can specify DIST=MULT on the MODEL statement to fit a multinomial model. The Getting Started example describes a lot of the syntax, although the example is for a logistic model (two classes) instead of your three-class multinomial model.