Positive marginal log-likelihood

[Prez Jarzebowski]

Dear all,

I am confused by positive values for marginal log-likelihood produced by some models. Am I wrong to assume these correspond to 0-1 values of likelihood so the log-likelihood should be negative? INLA Gitbook by Virgilio Gómez-Rubio gives examples of models with positive marginal log-likelihood, e.g. in 3.3.10 Seasonal random effects

I noticed models with positive log-likelihood give a negative DIC. Can DIC still be used the same way to compare the models with opposite DIC signs? 

Please forgive me these basic questions but I'm new to INLA.

Kind regards,

Prez

[INLA help]

Its log marginal likelihood, yes. 

—

Haavard Rue

Helpdesk

he...@r-inla.org

--
You received this message because you are subscribed to the Google Groups "R-inla discussion group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to r-inla-discussion...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/r-inla-discussion-group/0087118b-e772-4d1e-bde1-67725ee73648n%40googlegroups.com.

[Finn Lindgren]

The log-likelihood can be positive for continuous outcomes; the 3.3.10 example has a Gaussian conditional distribution as the observation likelihood.

Finn

--
You received this message because you are subscribed to the Google Groups "R-inla discussion group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to r-inla-discussion...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/r-inla-discussion-group/0087118b-e772-4d1e-bde1-67725ee73648n%40googlegroups.com.

--

Finn Lindgren
email: finn.l...@gmail.com

[Przemek Jarzębowski]

My confusion cleared, thank you both for the answer!

--

Przemek Jarzębowski