Getting clear on failure rate, the exponential distribution, $\beta$, and so on.

77 Views Asked by At

As I understand it the failure rate is the number of failures per unit time. I'm reading a book in which a collection of independent random variables $X_1, X_2, ...$ are assumed to represent the time until failure of the $i$th component, and are assumed to be exponentially distributed with parameter $\theta$. So $X_i$ has pdf

$$f(x)=\theta e^{-\theta x}$$

Since $E[X_i]=1/\theta$ then this represents the mean time until first failure, and $\theta$ represents the failure rate, the number of failures per unit time. However, the book writes that we assume there are about 0.5 failures per year, so I would think $\theta=0.5$. However, it goes on to say this is the same as a $\Gamma$-distribution with $\alpha=1$ and $\beta=2$. We define the $\Gamma$-distribution as

$$g(x)=\frac{\beta^\alpha}{\Gamma(\alpha)}x^{\alpha-1}e^{-\beta x}$$

That makes me think $\theta=\beta=2$ ... so I'm now confused about what the meanings of these parameters are.

If it's useful, I'm reading Degroot and Schervish's Probability and Statistics Fourth Ed., chapter 7, example 7.1.1.

1

There are 1 best solutions below

2
On BEST ANSWER

In this specific case, I think you are not understanding the example correctly. The book does not say that these two are the same distributions. It says that given $\theta$, the samples $X_i\sim\mathrm{Exp}(\theta)$ which has the pdf $f_X(\theta)=\theta e^{-\theta x}$ as you mentioned.

What is says next, is that there is a prior belief about the parameter $\theta,$ which is that $\theta$ is around $0.5$. An example of such a prior is the Gamma$(1,2).$ Indeed, this Gamma distribution has mean $1/2=0.5$ and hence supports the prior belief.