1
$\begingroup$

For the class $\{F_{\theta_1}, F_{\theta_2}\}$ of two DFs where $F_{\theta_1}$ is $N(0,1)$ and $F_{\theta_2}$ is $C(0,1)$, find a sufficient statistic.

Let, $X_1, X_2, \dots, X_n$ is a random sample from the class $\{F_{\theta_1}, F_{\theta_2}\}$.

Let, $Y_i$ is $1$ if $X_i$ is from $F_{\theta_1}$ and is $0$ if $X_i$ is from $F_{\theta_2}$.

Then the joint pdf of $x = (x_1,x_2,\dots,x_n)$ given $y = (y_1,y_2,\dots, y_n)$ is:

$$f(x|y) = \prod_{i=1}^{n}(f_{\theta_1}(x_i))^{y_i}(f_{\theta_2}(x_i))^{1-y_i}$$

where $f_{\theta_1}$ is pdf of $N0,1)$ and $f_{\theta_2}$ is pdf of $C(0,1)$.

(Note that $N(0,1)$ is standard normal distribution and $C(0,1)$ is standard Cauchy distribution.)

I don't know how to find sufficient statistic in this case (I don't even know if I am doing in right way). Thanks for any help.

$\endgroup$
1
$\begingroup$

This might work :

Define $$I(\theta)=\begin{cases}1&,\text{ if }\theta=\theta_1 \\ 0&,\text{ if }\theta=\theta_2\end{cases}$$

Then pdf of $X\sim F_{\theta}$ can be written as

\begin{align} G_{\theta}(x)&=[f_{\theta_1}(x)]^{I(\theta)}\,[f_{\theta_2}(x)]^{1-I(\theta)} \\\\&=\underbrace{\left[\frac{f_{\theta_1}(x)}{f_{\theta_2}(x)}\right]^{I(\theta)}}_{g(\theta,T(x))}\,\underbrace{f_{\theta_2}(x)}_{h(x)}\qquad,\,\,\theta\in\{\theta_1,\theta_2\} \end{align}

By Factorisation theorem, a sufficient statistic for $F_{\theta}$ is $$T(X)=\frac{f_{\theta_1}(X)}{f_{\theta_2}(X)}$$

$\endgroup$

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.