Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

Clarication Poisson Process and Poisson Distribution

17th July 2013

Poisson Distribution: First let us discuss about Poisson distribution and then we will discuss about Poisson process. Let X be a random variable that takes discrete values 0, 1, 2, ....... If X follows Poisson distribution then probability mass function of X or in other words probability that X takes value x is given as follows: P (X = x) =
e1 x 1 x! ; x

= 0, 1, 2, .....; 1 > 0

We write X P oisson(1 ) where 1 is parameter of the Poisson distribution. We also have E (X ) = 1 and V ar(X ) = 1 . This is the same 1 as the parameter of the distribution. Now let us look at question 52 of Tutorial sheet 4 that we discussed in class yesterday (16th July2013). X : random variable denoting the number of errors per page in a book. It is given that X P oisson(1 ). It is also mentioned that in 590 pages there are 390 typo errors. So av39 erage or mean number of errors per page = 59 . Since mean for Poisson is 39 1 we get, 1 = 59 . Probability of no error in a page = P (X = 0) = e1 = e 59 .
39

So the nal question that needs to be answered is that you are given 5 pages from the book. What is the probability of no errors on these pages ? Clarication 1: The clarication that I want to provide here is that we are not selecting the pages from the book. We have been given 5 pages and we need to give the probability of no errors in these 5 pages. So in this scenario the nal answer should be (e 59 )5 = e 59 5 . Clarication 2: We are using probability of number of errors per page because it is given in question that number of errors per page follows Poisson distribution. Clarication 3: One more doubt was whether the Poisson approximation to binomial can be used. Only if you are given a scenario where the number of trials (n) and probability of success in each trial (p) is given with n large and p small (refer to Sirs slides for the range of values)you can use Poisson approximation. If it is directly mentioned in the question that random variable follows Poisson distribution use this information to answer. Poisson Process: Now let us move onto Poisson Process. Let us consider the example that we took in class yesterday i.e. customers coming to a bank counter at random time points. We consider a process that counts the number of such arrivals (also called events) in any arbitrary time interval. We also know that the process follows the following properties. (i) events occurring in disjoint intervals are independent. (ii) in a small interval of time (t, t + t), the probability of more than one event occurring is negligible. (iii) the probability of one event occurring in (t, t + t) is t (here is intensity of the process and it is a positive constant). Under these conditions we can show that:
39 39

(a) the number of events occurring in any interval of length t is a random variable X where X P oisson(t). (b) the time between two events i.e. inter arrival time or waiting time (mentioned in Sirs slide)is a random variable Y where Y exp(). Clarication 4:Note in class I had mentioned the parameter of distribution of X as . This should be t. Clarication 5: When we talk of a Poisson process which counts number of events in a time interval t, the concerned Poisson distribution has the parameter t. This t is referred to as in Sirs slides. Also if you compare with 1 that we used for Poisson distribution (at the beginning of this document) we have 1 = t. Question 55 in the tutorial sheet 4 is an example of Poisson process. Here = 8 complaints / day. t = 1 since we are counting number of complaints in a day. Hence = t = 8 1 = 8.

You might also like