This is a General Linear Models topic but I believe it's just basic failure to remember some more basic math rules that's making it difficult for me.

If the link function is

$$g(\pi) = \log(\frac{\pi}{1-\pi}) = x^T\beta$$

show that this is equivalent to modeling the probability $\pi$ as,

$$\pi = \frac{e^{x^T}\beta}{1+e^{x^T}\beta}$$

Again I think this is a simply inverse. But I'm not quite sure how to do it.


You're right, it is just the inverse

$$ \ln\frac{\pi}{1 - \pi} = x^T\beta ~~~\Rightarrow~~~ \frac{\pi}{1 - \pi} = e^{x^T\beta} ~~~\Rightarrow~~~ \pi = (1 - \pi)e^{x^T\beta} = e^{x^T\beta} - \pi e^{x^T\beta} $$

Now rearrange a bit things

$$ \pi (1 + e^{x^T\beta}) = e^{x^T\beta} $$

and from here

$$ \pi = \frac{e^{x^T\beta}}{1 + e^{x^T\beta}} $$

  • $\begingroup$ Thank you kindly! I'm not sure why I kept stopping myself at the part where we multiply the exponential by $(1-\pi)$. I appreciate your time! $\endgroup$ – Nicklovn Feb 4 at 19:24
  • $\begingroup$ @Nicklovn Happy to help $\endgroup$ – caverac Feb 4 at 19:25

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.