I started to play with you very nice library and I was surprised by to get a positive log likelihood
I used as basis an harmonic model with 3 sin and cos pairs to model a timeseries, while I left the uninformative prior. How it is possible to get positive loglike?
Maybe because the marginal likelihood p(Data|Model), or just the likelihood p(Data|Hypothesis, Model), is not a probability, it is a density that can be greater than 1.
I started to play with you very nice library and I was surprised by to get a positive log likelihood
I used as basis an harmonic model with 3 sin and cos pairs to model a timeseries, while I left the uninformative prior. How it is possible to get positive loglike?