M.L.E. of binomial distribution

204 Views Asked by At

I have two questions.

1) In maximum likelihood estimation of binomial distribution I did as follows :
$L(p)=$$\prod$$ n\choose x_i $$p^{xi}(1-p)^{n-x_i}$
=$ n\choose x_1 $$n\choose x_2 $...$ n\choose x_n $ $p^{\sum x_i}(1-p)^{(n-x_1)+(n-x_2)+...(n-x_n)}$.

My question isn't $(1-p)^{(n-x_1)+(n-x_2)+...(n-x_n)}=(1-p)^{n^2-\sum x_i}$.

Then the M.L.E I get is =$\bar x\over n$ .
Is this wrong?

The article here says $\hat{p}=\frac{\sum_{i=1}^{n}{x}_{i}}{n}=\frac{k}{n}$.

Here why is it written as $\frac{\sum_{i=1}^{n}{x}_{i}}{n}$ and not as $\bar x$. ?

2)Here does ${\sum_{i=1}^{n}{x}_{i}}$ refer to the number of successes in n independent trials?
What is meant by number of successes in n independent trials?

> y<-rbinom(20,20,0.4)
> y
 [1] 11  3  5  8 10  9  6 11  7  9  8  5  6  6  6  9  8 10  4  7
> x<-sum(y)
> x
[1] 148

Does it mean if I generate some random numbers as then rbinom gives the number of successes in each trial so ${\sum_{i=1}^{n}{x}_{i}}$ refers to sum of rbinom values. Is the way I have understood ${\sum_{i=1}^{n}{x}_{i}}$ correct ?