i stumbled upon the following formula that describes making predictions based on the MAP estimate. i understand that we "plug in" in the MAP estimate in the predictive posterior, but i do not understand how we get from the second line to the third line, and why we lose the integral and the second term. could anyone help me understand this? Many thanks in advanceenter image description here

  • 1
    $\begingroup$ Could you at least cite where you stumbled on this calculation, and provide some context for what the various symbols are? $\endgroup$ – Dilip Sarwate Jul 10 '15 at 12:56
  • $\begingroup$ I guess it is from the tutorial "ML, MAP, and Bayesian — The Holy Trinity of Parameter Estimation and Data Prediction" by Avinash Kak from Purdue University. $\endgroup$ – groove Jun 27 '18 at 5:14

Addressing only the question you asked, it's because once you plug in the MAP estimate the first probability no longer depends on theta. Since it doesn't (and the integral is in terms of theta) you can pull it out of the integral and the resulting integral goes to 1, leaving you with the third line.

  • $\begingroup$ Why do we plug it only in the first probability? Shouldn't we plug it in both two probabilities and integrate with respect to it? $\endgroup$ – groove Jun 27 '18 at 5:12

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.