The random variable X has a Poisson distribution with unknown mean λ, where 0 < λ < ∞.

Based on a single X , Is it possible to calculate an unbiased estimator of $ e^{−2λ}$ ?

I have taken an Indicator random variable as follows :-

$I_A = P(X=0) \qquad\text{if } X=0\\ \,\,\,\,\,\,= 0 \quad\qquad\qquad\text{if } X\neq0 $

Then $E(I_A) = e^{-2\lambda}$

However I am not sure if this is correct .IF This is not correct kindly help me out.


Ps: I have found a similar question , :Finding an unbiased estimator of $e^{-2\lambda}$ for Poisson distribution

  • 3
    $\begingroup$ There is only one unbiased estimator of $e^{-2 \lambda}$ from a single observation, namely $(-1)^X$, and it is not a particularly sensible estimator as it only gives the values $+1$ or $-1$ when in fact $0 \lt e^{-2 \lambda} \lt 1$ unless $\lambda=0$. See en.wikipedia.org/wiki/… $\endgroup$ – Henry Sep 10 '18 at 7:43

Your proposed estimator is not $I_A$ for any $A$. If it was, it should return $1$ when $X\in A$ and $0$ otherwise.

Moreover, it is also wrong to say that $$\hat \theta =\left\{\begin{matrix}P(X=0)&X=0,\\0&X\neq0,\\\end{matrix}\right.$$ is an estimator for $\theta=e^{-2\lambda}$, since $P(X=0)=e^{-\lambda}=\sqrt\theta$. Your proposed estimator for $\theta$ can be anything BUT a function that depends on $\theta$ itself.


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.