Log Likelihood Positives in Logit Models

[Muhammad Hafiz Aditya]

Hi Michel and all,

I am trying to estimate parameters for a MNL model. The model I'm trying to estimate is so simple, it has 2 parameters and 24 observations. 

I'm just want to compare Biogeme's result with the result from Apollo. The results are different, and i notice something weird that the log likelihood from biogeme are positive, both init and final.

Then i go to your example that uses swiss metro data and try to change a little. I do a simple tweak to change the models from loglogit to logit, then i also get positive log likelihood from the results. 

Would really help if u tell me where i went wrong!

Thanks in advance!

[Bierlaire Michel]

Use the online examples as a reference. https://github.com/michelbierlaire/biogeme/blob/master/examples/swissmetro/01logit.py
Or watch the video https://youtu.be/vS-Sg0htQP4 (although this was made for a previous version of the software)

> --
> You received this message because you are subscribed to the Google Groups "Biogeme" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to biogeme+u...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/biogeme/077b869c-60e5-46e1-8cbc-213b8f8f7f8fn%40googlegroups.com.

Michel Bierlaire
Transport and Mobility Laboratory
School of Architecture, Civil and Environmental Engineering
EPFL - Ecole Polytechnique Fédérale de Lausanne
http://transp-or.epfl.ch
http://people.epfl.ch/michel.bierlaire

[Muhammad Hafiz Aditya]

Okey thank you!

When i used loglogit models, the output parameter like LL, AIC, BIC are same for Biogeme and other package. But why are the beta values different, and the difference is huge! Can you enlighten me on where i went wrong?

Thank you.