Each usable run with its confidence and anyApplies calibrated out of fold: the cases are dealt
into folds (AssignCaseFolds), and each fold's runs are calibrated by two Platt fits on the
other folds' runs only (the confidence on their agent runs). No run is scored by a fit that saw
it. Null with fewer than two cases.
Each usable run with its confidence and
anyAppliescalibrated out of fold: the cases are dealt into folds (AssignCaseFolds), and each fold's runs are calibrated by two Platt fits on the other folds' runs only (the confidence on theiragentruns). No run is scored by a fit that saw it. Null with fewer than two cases.