3
$\begingroup$

For the random variable X with a log-normal pdf

$f(x)=\frac{1}{\sqrt{2*\pi}}x^{-1}e^{-.5*log(x)^2}$

I am trying to find a location-scale family $h(x)$ such that $h(x)$ has mean 0 and variance 1

Now I know for location-scale family the transformation

$h(x)=\frac{1}{\sigma}f(\frac{x-\mu}{\sigma})$

I found the expected value and variance of $f(x)$ respectively to be $\sqrt{e}$ and $(e-1)e$ but is it just a matter of plugging those in to the above to find what I need? Just a bit confused about what the transformation does. Thanks in advance

$\endgroup$
  • 1
    $\begingroup$ Is this for study purposes, or do you need it for some task? The idea of a location-scale family is not that it has mean 0 and variance 1 (that's one member, not a family). Can you give a definition of $h$ (that isn't contradictory)? You might base the family off a standardized base though, for convenience of interpretation. The usual lognormal is already a scale-shape family, so you'd just need to add a location shift to that. Your formulation assumes a particular shape parameter -- did you want to restrict yourself to only that shape? $\endgroup$ – Glen_b -Reinstate Monica Feb 4 '15 at 23:32
  • $\begingroup$ Yeah this is a problem in one of my textbooks that I'm trying to work on mostly for study purposes and that was the problem given. I suppose I might just be misunderstanding what it's asking for. $\endgroup$ – James Snyder Feb 5 '15 at 1:07
  • 1
    $\begingroup$ Thanks. [Could you add the self-study tag and see the guidelines in its tag wiki? You may need to add some details to follow the guidelines.] Does the book give a definition of $h$ somewhere else, perhaps when it first mentions location-scale families? $\endgroup$ – Glen_b -Reinstate Monica Feb 5 '15 at 1:46
1
$\begingroup$

I don't think you need to worry about the mean and variance of $f$ as such.

It gives you $f$ and it gives you the transformation (the transformation isn't what you said, the transformation is $Y=\frac{X-\mu}{\sigma}$); $h$ is actually the density of the transformed random variable.

[Presumably you know how to use the usual methods to do the transformation; the Jacobian is just $\frac{1}{\sigma}$. Have you seen the result that for $Y=g(X)$, the density is $f_Y(y)= f_X(g^{-1}(y)) \left|\frac{d g^{-1}(y)}{dy}\right|$? )

So anyway, in the question, they've given you the result of doing that transformation already, it's really just a matter of direct substitution from there.

$\endgroup$

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.