Could someone explain this concept or give a link to the explanation of this concept please? I know what "Bayesian" is, but I don't know what "Bayesian evidence" is. A good explanation of "evidence" will also be appreciated.

  • $\begingroup$ Not sure exactly what is meant by "Bayesian Evidence." I don't really think this is a thing. A quick search suggested these topics might be related: en.wikipedia.org/wiki/Evidence_under_Bayes_theorem , en.wikipedia.org/wiki/Bayesian_inference (I could imagine someone mispronouncing 'inference' as 'evidence'.) $\endgroup$ – Newb Oct 29 '13 at 5:54
  • $\begingroup$ @Newb Thanks for the reply. I think hejseb is right. This term shows up in a paper for several times and I'm sure that this is not a misspell. Maybe this is just an informal way to say it among non-mathematicians (I'm not sure). $\endgroup$ – xuhdev Oct 30 '13 at 22:16

This is a concept used in Bayesian model selection. Essentially, it's the probability of the data in some model $\mathcal{M}$ when the parameter uncertainty has been integrated out. Mathematically, it is

$$ Pr(D|\mathcal{M})=\int_{\Theta}Pr(D|\theta)Pr(\theta|\mathcal{M})d\theta $$ where $\Theta$ is the domain of $\theta$. Using the evidence of different models you can compute the Bayes factor, for example.

See this for further reading.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.