What is the best intuition behind the unique parameter $\lambda$ in the Poisson distribution?
$\begingroup$
$\endgroup$
Add a comment
|
2 Answers
$\begingroup$
$\endgroup$
6
Poisson RV is commonly used for modelling number of occurrences of an event within a particular time interval. And, since $E[X]=\lambda$, its unique parameter is referred as mean number of event occurrences within our particular time interval.
-
$\begingroup$ That's nice. Can be something be added about the "particular time inetrval" ? $\endgroup$ Oct 21, 2018 at 12:29
-
$\begingroup$ e.g. $X$: number of customers arriving daily; here time interval is day. $\endgroup$– gunesOct 21, 2018 at 12:31
-
$\begingroup$ Is the Poisson distribution strictly speaking discrete or can it be even continuous ? I.e. can we say that there is a probability of number customers arrving daily 2.45 ? $\endgroup$ Oct 21, 2018 at 12:42
-
$\begingroup$ Poisson is a discrete RV. It can be 0,1,2,... $\endgroup$– gunesOct 21, 2018 at 12:48
-
$\begingroup$ Yes but it has continuous analog if we intersperse curve through the discrete values at $0,1,2,...$ Can this be done and understood formally? $\endgroup$ Oct 21, 2018 at 13:00
$\begingroup$
$\endgroup$
Since the Poisson distribution is defined as
$$\pi_{\lambda}(k)~=~\frac{\lambda^k}{k!}e^{-\lambda}$$
It is not hard to show that $E[X]=\lambda$ and $D^2[X]=\lambda$. Therefore you can directly interpret $\lambda$ as the expectation or as well as the variance of the poisson distributed variable.
