One question's check: passes when the model picked right, and carries
its probability measures (brier, logLoss, confidence, correct), which a
test averages and turns into a calibration error. Doesn't apply (null)
when there's no right answer to compare with.
One question's check: passes when the model picked right, and carries its probability measures (brier, logLoss, confidence, correct), which a test averages and turns into a calibration error. Doesn't apply (null) when there's no right answer to compare with.