0
$\begingroup$

Let $N_t$ denote the number of accidents upto time $t$. Assume that $ N_t$ is Poisson process with intensity 2. Given that there are exactly 5 accidents in the period $[20,30]$ what is the probability that there is exactly one accident during the time period $[15,25]$

Solution


$[15,25]_1 \longrightarrow $ event of exactly one accident in $[15,25]$ time period

$[25,30]_4 \longrightarrow $ event of exactly 4 accidents in $[25,30]$ time period

$[20,30]_5 \longrightarrow $ event of exactly 5 accidents in $[20,30]$ time period

So we need to calculate $ P([15,25]_1)$ given $[20,30]_5$. So using Bays theorem the probability is

\begin{align} \frac{P \left([20,30]_5 \cap [15,25]_1\right)}{P([20,30]_5)} & = \frac{P([15,25]_1 \cap [25,30]_4)}{P([20,30]_5)}\\ & = \frac{P([15,25]_1) \cap P([25,30]_4)}{P([20,30]_5)}\\ & = \frac{\frac{5^1e^{-5}}{1!}\frac{2.5^4e^{-2.5}}{4!}}{\frac{5^5e^{-5}}{5!}} \end{align}

I am using the fact that intensity is 2 (i.e an accident every 2 minutes), we have that $ \lambda =0.5$ (accidents per minute).

This calculation leads to the final answer $\frac{5}{16}e^{-2.5} $. But the answer provided is $\frac{15}{32}e^{-10}$ where did I go wrong?

$\endgroup$
2
$\begingroup$

It appears you have not taken into account the time elapsed in each interval. Specifically, if the intensity is $\lambda = 2$ per unit of time, then $$\Pr[N_t = n] = e^{-\lambda t} \frac{(\lambda t)^n}{n!}, \quad n = 0, 1, 2, \ldots.$$

The other problem is that you need to take into account the fact that there are two cases to consider: either the event that occurs in $(15,25]$ occurs in $(20,25]$, or it occurs in $(15,20]$. In the first case, then we have $0$ events in $(15,20]$, $1$ event in $(20,25]$, and $4$ events in $(25,30]$. In the second case, we have $1$ event in $(15,20]$, $0$ events in $(20,25]$, and $5$ events in $(25,30]$. To obtain the total probability of the desired event, you need to consider these two mutually exclusive cases.

$\endgroup$
  • $\begingroup$ +1 Exactly. And disjointness of intervals guarantees independence of the corresponding events. $\endgroup$ – drhab Dec 6 '17 at 9:36
0
$\begingroup$

Here is the completed solution just in case, you are still wondering.

enter image description here

$\endgroup$

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.