What is the log likelihood used for?


Computing the log likelihood is taking a long time a few of my models. Can anyone tell me what it’s used for and if I set the parameter to not compute the log likelihood, what do I lose in terms of my analysis of the model after? I don’t think I use it now unless I am in the background without realizing it.

It’s used for model comparison when calling arviz.compare. You can disable it (it’s actually disabled by default in more recent versions of PyMC)