multinomial logit model in rstanarm

[malai...@gmail.com]

Hi All,

Just a quick question, is it possible I can the change the logit model to multinomial logit model in rstanarm? or is there a way I can fit my data using the multinomial logit model instead of logit model in rstanarm?

Thanks 

stan_glm1 <- stan_glm(formula = y ~ Intersection_Type + Driver1_Age +Driver1_Gender + Driver2_Age + Driver2_Gender + 

                        Crash_Type + Driver_Turning +Atfault_Lane_Position + Travel_Street + Approach_Through_Lanes +

                        Approach_Left_Turn_Lanes + Approach_Right_Turn_Lanes +Approach_Signal_Type + Crash_Day + 

                        Crash_Time + surfacewidth + medianwidth + AADT + RCI_AVGTFCT + cat_speed + Lighting_Condition +

                        Weather_Condition + Road_Surface_Condition + mediantype +  workzone,

                      data = agingcrashes, 

                      family = binomial('logit'), 

                      chains = 4, 

                      seed = 123, 

                      warmup=2000, 

                      iter = 5000,

                      prior = normal(0,1),

                      prior_intercept = cauchy(0,2.5),

                      QR= TRUE)

[Daniel Guhl]

Hi,

brms can fit multinomial logit models using Stan, see https://cran.r-project.org/web/packages/brms/vignettes/brms.pdf

Best, 

Daniel

--
You received this message because you are subscribed to the Google Groups "Stan users mailing list" group.
To unsubscribe from this group and stop receiving emails from it, send an email to stan-users+...@googlegroups.com.
To post to this group, send email to stan-...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[malai...@gmail.com]

Thanks, let me check that now.

[Ben Goodrich]

On Tuesday, October 25, 2016 at 2:26:08 PM UTC-4, dg wrote:

brms can fit multinomial logit models using Stan, see https://cran.r-project.org/web/packages/brms/vignettes/brms.pdf

The rstanarm package doesn't have a multinomial logit implemented. We are working on a multinomial probit.

Ben

[Stephen Martin]

Is there a reason you're implementing multinomial probit instead of multinomial logit?
Just curious, because I fit multinomial logit via stan (using the softmax function with one group's predicted value being set to 0 for scale).

[Ben Goodrich]

On Sunday, November 20, 2016 at 11:05:21 PM UTC-5, Stephen Martin wrote:

Is there a reason you're implementing multinomial probit instead of multinomial logit?
Just curious, because I fit multinomial logit via stan (using the softmax function with one group's predicted value being set to 0 for scale).

The multinomial logit would certainly be easier but it is hard to get behind the assumption that the errors are independent across choices.