1
$\begingroup$

Let $X, Y$, and $Z$ be three independent Poisson Y3 random variables with parameters $\lambda_1, \lambda_2, \lambda_3$, respectively. For $y=0, 1, 2,\ldots, t$, calculate:

$$P(Y = y \mid X + Y + Z = t).$$

Unsure of how to approach this. Given formulas from my book, I get $$e^{-3}\cdot \frac{3^t}{t!}$$

$\endgroup$
0
$\begingroup$

\begin{align} & \Pr(Y=y \mid X+Y+Z = t) = \frac{\Pr(Y=y\ \&\ X+Y+Z = t)}{\Pr( X+Y+Z = t )} \\[10pt] = {} & \frac{\Pr(Y=y\ \&\ X+Z = t-y)}{\Pr( X+Y+Z = t )} = \frac{\Pr(Y=y)\cdot\Pr(X+Z = t-y)}{\Pr( X+Y+Z = t )} \tag 1 \end{align} Do you know that the sum of independent Poisson-distributed random variables also has a Poisson distribution? If so, you get $X+Y+Z\sim\mathrm{Poisson}(\lambda_1+\lambda_2+\lambda_3)$ and $X+Z\sim\mathrm{Poisson}(\lambda_1+\lambda_3)$ and $Y\sim\mathrm{Poisson}(\lambda_2)$, so plug those into the last expression in $(1)$ and simplify. Every occurence of the number $e$ will cancel out. You will find that the conditional distribution of $Y$ is a binomial distribution.

$\endgroup$
  • $\begingroup$ thank you for this explanation @Michael. So Pr(Y=y) would be e^-lambda2* lambda2^y/y! ? Not entirely certain about Pr(X+Z=t-y) $\endgroup$ – samp1920 Jul 14 '15 at 1:47
  • $\begingroup$ @samp1920 : Same sort of thing: $\displaystyle \Pr(X+Z=w) = \frac{(\lambda_1+\lambda_3)^w e^{-(\lambda_1+\lambda_3)}}{w!}$. ${}\qquad{}$ $\endgroup$ – Michael Hardy Jul 14 '15 at 2:52

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.