2
$\begingroup$

I feel like this is something we went over in class but it's not coming to me for some reason. I need to find the Bayesian estimator for $\tau(\theta)=e^{-\theta}$ under square error loss. I already found the Bayesian estimator for $\theta$ (idk if you'll need the context but the prior was Gamma($\alpha,\beta$) and the parameter of interest was that of iid Poisson variables). I don't know if I can just take the expectation of $e^{-\theta}$ and call it a day or what I need to do. Any help would be appreciated.

$\endgroup$
2
  • 5
    $\begingroup$ No invariance property. Bayes estimator of $\theta$ under squared error loss is $E[\theta\mid X]$. Similarly Bayes estimator of $\tau(\theta)$ under squared error loss is $E[\tau(\theta)\mid X]$. $\endgroup$ Commented Apr 12, 2020 at 7:43
  • 3
    $\begingroup$ If $\exp\{-\theta\}$ is estimated under the squared error loss, the Bayes estimator is the posterior expectation. It is unclear what you mean by invariance in this context. $\endgroup$
    – Xi'an
    Commented Apr 12, 2020 at 9:43

1 Answer 1

0
$\begingroup$

Thanks to the people in the comments, I can conclude that Bayesian estimators under square error loss do not have an invariance property, and the Bayes estimator of $\tau(\theta)$ under square error loss is $E(\tau(\theta)|X)$.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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