Maximum Likelihood estimation
<p>Oct 12, 2014 08:57 PM exclt videos by Ben Lambert (UK) 3 parts. all excellent; watch together;</p>
<p>
population: whole of the UK 70 M people;</p>
<p>
some value of p (prob of being male);
we only have a sample: all we have is an estimate p hat of p given our sample.</p>
<p>
L = likelihood = product of probs assuming it is p.</p>
<p>
dL/dp = 0... need to use product rule to differentiate.</p>
<p>
but take l = little l = log L = log product...</p>
<p>
get same estimate; log is monotonic;</p>
<p>
log ab = log a + log b;</p>
<p>
so, get SUM log(p to the xi etc)</p>
<p>
p hat = x bar; = sum of xi divide by N
just the fraction of males in the sample.</p>
<p>
</p>
<p>
</p>