What is meant by the inverse of a CDF? Logit vs. Logistic Regression.

19 Views Asked by At

There is seemingly a difference whether one approaches the "logit" from statistics / econometrics:

$$F^{-1}(\pi) = \log\left(\frac{\pi}{1-\pi}\right) = \text{logit}(\pi)$$ (I)

Or from the machine learning perspective, where it is typically called "logistic regression":

$$F(\eta) = \frac{\exp(\eta)}{1 + \exp(\eta)} = \frac{1}{1 + \exp(-\eta)}$$ (II)

There is essentially no difference between what is done to classify new observations. But what exactly does that mean the inverse of the CDF? Is there a possibility to reformulate (I) to get to (II)?

Thanks!