Interview Guides
You have a model that outputs probabilities, and the team wants to know whether those scores mean what they say. The model may rank cases well, but the question is whether a predicted 0.80 really corresponds to about an 80% event rate.
How would you measure whether a model is calibrated?
