0
$\begingroup$

The output of a certain integrated-circuit production line is checked daily by inspecting a sample of $100$ units. Over a long period of time, the process has maintained a yield of $80$ percent, that is, a proportion defective of $20$ percent, and the variation of the proportion defective from day to day is measured by a standard deviation of $0.04$.
If on a certain day the sample contains $38$ defectives, find the mean of the posterior distribution of $\theta$ as an estimate of that day’s proportion defective. Assume that the prior distribution of $\theta$ is a Beta distribution.

I know $\textbf{posterior} \propto \textbf{likelihood} \times \textbf{prior}$, and it seems like the likelihood function is binomial so $$f(x|\theta)\propto \theta ^x(1-\theta)^{n-x} $$ and the prior is $$h(\theta )\propto \theta^{\alpha - 1} (1- \theta)^{\beta - 1}~,$$ right? It seems like the posterior is supposed to look like a beta distribution, but where do the values $0.04$ and $38$ come into play here?

$\endgroup$
0
$\begingroup$

On this certain day the Binomial as you stated has $n = 100$ and $x = 38$.

The Beta prior as you stated $h(\theta )\propto \theta^{\alpha - 1} (1- \theta)^{\beta - 1}$ has parameters such that

  • the mean is $\dfrac{ \alpha }{ \alpha + \beta } = 0.2$

  • the variance is $\dfrac{ \alpha \beta }{ (\alpha + \beta)^2 (\alpha + \beta + 1) } = 0.04$

which determines uniquely the values to be $~\alpha = \frac35$ and $\beta = \frac{12}5$.

$\endgroup$

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.