i've looked everywhere I could for an answer to this question but no luck ! If I have $X_1 .... X_n$ random variables that are independent and identically distributed such as ∀ $1 <$ $i$ $<n$, $X_i$ ~ B(n,$\theta$) (binomial distribution)

I know that the likelihood is :

$P_n$($\theta$,x)=$\prod\limits_{i}$ $\binom{n}{x_i}p^{x_i}(1-p)^{n-x_i}$

but then it seems kind of hard to calculate as product, I tried to calculate log($p_n$) but then the $x_i$! causes me problems

Can you enlighten me ? what is the max-likelihood of a binomial distribution ?


Take the log-likelihood function, i.e. $$L(p) = \log \prod_i \binom{n}{x_i} p^{x_i}(1-p)^{n-x_i}$$ which becomes $$L(p) = \sum_i \log\binom{n}{x_i} p^{x_i}(1-p)^{n-x_i}$$ even more $$L(p) = \sum_i \log\binom{n}{x_i} + \sum_i x_i\log p + \sum_i(n-x_i)\log(1-p)$$ Since you're interested in the ML estimate of $p$. let's derive with respect to $p$ and set it to zero. $$\frac{d}{dp}L(p) = \frac{1}{p}\sum_i x_i - \frac{1}{1-p}\sum_i(n-x_i) = 0$$ which gives you $$\frac{1}{1-p}\sum_i(n-x_i) = \frac{1}{p}\sum_i x_i$$ or $$\frac{\sum_i(n-x_i)}{\sum_i x_i} = \frac{1-p}{p} = \frac{1}{p} - 1$$ Now find $p$. $$\frac{1}{p} = \frac{\sum_i(n-x_i) + \sum_i x_i}{\sum_i x_i} = \frac{\sum_i n}{\sum_i x_i} $$

  • $\begingroup$ thank you, it was just a matter of concentration and calculating step by step $\endgroup$ – Blueberry Feb 1 '19 at 16:03

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.