2
$\begingroup$

Im trying to prepare for an university exam in Probability, but I got stuck on this practice sheet question:

A company produces lots of highly complex devices. As a consequence, these devices may not always work properly. Suppose, individual devices are defective with a probability p, and these problems occur independently. Also the number X of devices produced on a certain day is random. It follows a Poisson probability law: $$ P[X = k] = \frac{\mu^k}{k!}e^{-\mu} ; k\ge0, \mu>0 $$ On a sunny day in October, how large is the probability to have no defective device at all?

My take on it was to define a new random variable Y as the number of deffective devices out of $k$ produced devices.Then the probability to get $n$ deffective devices out of $k$ produced would follow the binomial law, thus: $$P[Y = n] ={ k \choose n}p^n(1-p)^{k-n} $$ and for zero defective devices this gives us: $$P[Y = 0] =(1-p)^{k} $$ If X and Y are independent then the probability to get $0$ defective devices on a day is just going to be the product of those probabilities yielding: $$P[Y=0|X=k] = \frac{\mu^k}{k!}e^{-\mu}(1-p)^{k}$$ The problem is that now I am not sure whether X and Y are actually independent and also I couldn't come up with any way of checking whether my answer makes sense. Could someone please clarify and if I am wrong with my approach explain how to actually calculate the desired probability?

$\endgroup$
  • $\begingroup$ The random variable $Y$ that you introduce is a confusing one. It is connected with some fixed $k$ and just seems to "fall out of the sky". Is it defined on the same probability space as $X$? If yes, then how did you do that? If no then you cannot compare $X$ and $Y$ so that independence does not even has a meaning here. $\endgroup$ – drhab Jan 18 '18 at 15:31
0
$\begingroup$

Let $Z$ denote the number of defective devices on a day. Then actually you found that: $$P(Z=0\mid X=k)=(1-p)^k$$ and can go on with:$$P(Z=0)=\sum_{k=0}^{\infty}P(Z=0\mid X=k)P(X=k)=\sum_{k=0}^{\infty}(1-p)^ke^{-\mu}\frac{\mu^k}{k!}=e^{-\mu}e^{(1-p)\mu}=e^{-p\mu}$$

It can be shown that $Z$ has Poisson distribution with parameter $p\mu$.

$\endgroup$
  • $\begingroup$ Thank you for your answer! I just dont understand why does the summation go to infinity. If the number of produced devices for a day would be some finite value, lets say N, should it not go from 0 to N? $\endgroup$ – Stilian Jan 18 '18 at 18:00
  • $\begingroup$ If we say in our model that the number of produced devices on a sunny day in october is $N$ then $N$ is not a fixed number but a random variable. It can take values $0,1,2,3,\dots$, and if $N$ is assumed to have Poisson distribution then each of the events $\{N=0\},\{N=1\},\{N=2\}\dots$ has positive probability. $\endgroup$ – drhab Jan 18 '18 at 19:11

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.