2
$\begingroup$

I have this probability distribution:

$$P(\{X=x\})=\frac{\lambda^x\exp(-\lambda)}{x!(1-\exp(-\lambda))}$$

where $x\in \mathbb{N}$. If not for this factor $\frac{1}{1-\exp(-\lambda)}$, this would be Poisson distribution. What is its name?

$\endgroup$
  • 3
    $\begingroup$ For what values of $x$? Are you sure the probabilities sum to $1$? $\endgroup$ – Juho Kokkala Dec 14 '15 at 16:48
  • 1
    $\begingroup$ Hint: what values can $x$ take? $\endgroup$ – Scortchi - Reinstate Monica Dec 14 '15 at 16:48
  • $\begingroup$ @JuhoKokkala: Snap! $\endgroup$ – Scortchi - Reinstate Monica Dec 14 '15 at 16:48
  • 4
    $\begingroup$ You must specify the range of $x$! Without that, the definition is not precise. We cannot, for instance, decide if the probabilities sum to 1 or not. $\endgroup$ – kjetil b halvorsen Dec 14 '15 at 16:49
  • 4
    $\begingroup$ Where does $\mathbb{N}$ start for you? Nought or one? $\endgroup$ – Scortchi - Reinstate Monica Dec 14 '15 at 16:50
2
$\begingroup$

I have found the name. It's called "Zero-truncated Poisson distribution".

Source: Johnson L., Kemp A., Kotz S., Univariate Discrete Distributions 3rd edition, Wiley, p. 188

$\endgroup$
  • $\begingroup$ (+1) Note that $\exp(-\lambda) = \Pr(X=0)$ when $X$ is a Poisson random variable. So the mass function can be recognized as $\frac{\Pr(X=x)}{\Pr(X >0)}=\Pr(X=x \mid x>0)$, the Poisson probability of a count conditional on its being greater than nought. $\endgroup$ – Scortchi - Reinstate Monica Dec 15 '15 at 11:41

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.