topic Re: Compare goodness of fit between MNL and HB model - AIC, BIC, DIC, etc.? in Statistical Procedures

Compare goodness of fit between MNL and HB model - AIC, BIC, DIC, etc.?

ahoeweler — Mon, 14 Nov 2016 11:39:50 GMT

Hi everyone,

I want to compare the goodness of fit of a MNL model, that I conducted through proc phreg and ties=breslow and a HB model that I conducted through proc bchoice.

Which is the easiest way to compare the goodness of fit of those two models and how to I implement them in the codes?
I head that the Bayesian Information Criterion is useful but I couldn't find how you include it in the above named procedures.

Thanks for your help.

Best,

Alex

Re: Compare goodness of fit between MNL and HB model - AIC, BIC, DIC, etc.?

amyshi — Tue, 15 Nov 2016 15:49:42 GMT

There is DIC in BCHOICE that you can use to compare models.

Amy

Amy Shi

SAS

Re: Compare goodness of fit between MNL and HB model - AIC, BIC, DIC, etc.?

ahoeweler — Tue, 15 Nov 2016 16:14:27 GMT

Hi amyshi,

thanks for your answer, but how to I call the DIC measure in the MNL model that I call through the proc phreg command with the ties=breslow option? Because as far as I understood, the DIC measure is for Bayesian models but not for exclusive MNL models.

Best,

Alex

Re: Compare goodness of fit between MNL and HB model - AIC, BIC, DIC, etc.?

amyshi — Tue, 15 Nov 2016 16:28:24 GMT

BCHOICE fits both MNL with fixed effects only and HB MNL with random effects. The fixed effects only model should agree with PHREG. So just use BCHOICE for choice data.

I am part of the Bayesian Statistical Modeling group at SAS R&D that is responsible for developing the BCHOICE and MCMC procedures.

If you are interested and have more questions about BCHOICE, please write to me directly at amy.shi@sas.com

Thanks,

Amy

amy.shi@sas.com