In Bayesian Essentials, page 40, there's the following formula:

enter image description here

I've tried to derive it.

$P(M=k|D=\frac{P(D|M=k)P(M=k)}{\sum_jP(D|M=j)P(M=j)}=\frac{\int P(D|\theta,M=k)P(\theta|M=k)\ d\theta \ \ P(M=k)}{\sum_jP(D|M=j)P(M=j)}=\frac{\int l_k(\theta)\pi_k(\theta)\ d\theta \ \ P(M=k)}{\sum_jP(D|M=j)P(M=j)}$

However, I'm stuck at this last step...

Any help would be appreciated.


Sorry about this terrible typo!!!

$$ \mathbb{P}^\pi(\mathfrak{M}=k|\mathscr{D}_n) = \dfrac{\mathbb{P}^\pi(\mathfrak{M}=k) \int \ell(\theta_k|\mathscr{D}_n)\pi_k(\theta_k)\,\text{d}\theta_k}{ \sum_{j=1}^J \mathbb{P}^\pi(\mathfrak{M}=j)\pi_j(\theta_j)\,\text{d}\theta_j}\,. $$ should read as $$ \mathbb{P}^\pi(\mathfrak{M}=k|\mathscr{D}_n) = \dfrac{\mathbb{P}^\pi(\mathfrak{M}=k) \int \ell(\theta_k|\mathscr{D}_n)\pi_k(\theta_k)\,\text{d}\theta_k}{ \sum_{j=1}^J \mathbb{P}^\pi(\mathfrak{M}=j)\int \ell(\theta_j|\mathscr{D}_n)\pi_j(\theta_j)\,\text{d}\theta_j}\,. $$ which is also what you propose as the last equation in your question. For some reason the cut-and-paste between numerator and denominator did not work completely and we failed to spot the mistake. Apologies!

| cite | improve this answer | |
  • $\begingroup$ Thanks for the help. By the way, do you know of the existence of an errata for this book? $\endgroup$ – An old man in the sea. Jan 15 '17 at 19:45
  • $\begingroup$ @Anoldmaninthesea. There is no errata list so far but you demonstrated the need for one!!! $\endgroup$ – Xi'an Jan 26 '17 at 11:03

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.