S3E17: Logistic Regression: 2 Logit 2 Quit
Quantitude
00:00
The S Shape Algorithm of Probability
We have a non linear relation that we need to represent in our model. We don't even have probability as an outcome. As x gets higher and higher, our predictions should be either piking at or at zero. The s shape ally captures our belief about the way predictions of probabilities ought to behave. So where we're going to go is either thinking about fitting probabilities with an s shape, or taking those things and bending them to our will.
Transcript
Play full episode
