1
$\begingroup$

I have a binomial process with probability of success p. Let k be number of success in when the number of trails is N. I have defined s as the random variable of the share of success $\ s=\frac{k}{N}$

i need to find the limiting distribution of s as $\ \lim_{N\to \infty} s$

I have tried to go about it as follows. The chance of exactly k success is $C_k^N *p^k*(1-p)^(N-k)$

Then m=[N*s] where [] is the least integer function.The limit distribution will be the limiting value of $\ \lim_{N\to \infty}C_m^N *p^k*(1-p)^\left(N-m\right) $ I tried to draw some inferences using simulation.It seems that the limiting distribution of the share of success is a normal with mean p empirical distribution of share of success

$\endgroup$
4
  • $\begingroup$ The central limit theorem shows that the distribution actually is asymptotically normal $\endgroup$
    – Peter
    Commented Sep 9, 2017 at 19:24
  • 1
    $\begingroup$ You might be interested in reading about the De Moivre- Laplace theorem. $\endgroup$
    – kimchi lover
    Commented Sep 9, 2017 at 19:26
  • $\begingroup$ Hi Peter,what is the mean and variance of that distribtion? Just by eyeballing the distribution,it seems the mean is also p. $\endgroup$
    – kangkan Dc
    Commented Sep 9, 2017 at 19:48
  • $\begingroup$ @kangkanDc Yes, this makes sense. In the usual binomial distribution, the variance would be $np(1-p)$, but I am not sure whether we can use this result for this concrete distribution. Unfortunately, I have no idea what the variance is. $\endgroup$
    – Peter
    Commented Sep 9, 2017 at 21:08

0

Reset to default

You must log in to answer this question.