It is given that

$f_\eta(y) = h(y)exp(\eta T(y)-A^*(\eta))$

$P_Y(y)= \frac{k}{\lambda} (\frac{y}{\lambda})^{k-1}exp(-(\frac{y}{\lambda})^k)$

What i did was by arranging $P_Y(y)$ to get $\frac{k}{\lambda ^k} y^{k-1}exp(-(\frac{y}{\lambda})^k)$

so that $lnP_Y(y) = ln(\frac{k}{\lambda^k})+(k-1)ln(y)-(\frac{y}{\lambda})^k$


$T(y) = y^{k-1}$

$\eta = -(\frac{1}{\lambda})^k$


$A^* = -ln(\frac{k}{\lambda^k})= -ln(-k\eta)$

Am i correct?

and how do you work out the mean and variance of Y in terms of $A^*$?


For part one of your question:

No. In case you want to use this particular representation of exponential family [over a more relevant form $h(x)g(\eta) \exp(\eta T(x))$]

\begin{align*} f_\eta(y) &= h(y)\exp(\eta T(y)-A^*(\eta))\\ \ln f_{\eta}(y) &= \ln h(y) + \eta T(y) - A^*(\eta)\\ P_Y(y) &= \frac{k}{\lambda} (\frac{y}{\lambda})^{k-1}\exp(-(\frac{y}{\lambda})^k)\\ \ln P_Y(y) &= \ln y^{k-1} -(\frac{y}{\lambda})^k - \ln k - \ln \lambda^k\\ \ln P_Y(y) &= \ln ky^{k-1} + \eta (-y^k) - \ln \eta \text{ where } \eta = \lambda^k\\ \end{align*}


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.