eval_toolkit.calibration

DEFAULT_FN_COST	Convert a string or number to a floating-point number, if possible.
DEFAULT_FP_COST	Convert a string or number to a floating-point number, if possible.
DEFAULT_N_BINS	int([x]) -> integer int(x, base=10) -> integer
DEFAULT_PRIOR	Convert a string or number to a floating-point number, if possible.
DEFAULT_STRATEGY	str(object='') -> str str(bytes_or_buffer[, encoding[, errors]]) -> str
CostMatrix	Frozen scaffolding for FP/FN/abstain costs at an assumed prior.
bayes_optimal_threshold	Bayes-optimal threshold per Elkan 2001 [#elkan]_ cost-sensitive derivation.
fit_beta_calibrator	Beta calibration per Kull et al. 2017 [#kull]_.
fit_isotonic_calibrator	Niculescu-Mizil & Caruana 2005 [#nm05]_ isotonic regression.
fit_platt_calibrator	Platt 1999 [#platt]_ sigmoid scaling with Lin 2007 [#lin]_ Laplace-smoothed targets.
fit_temperature	Single-parameter temperature scaling per Guo et al. 2017 [#guo]_.
fit_temperature_binary	Binary-probability adapter for fit_temperature() (Guo et al. 2017 [#guo]_).
fit_temperature_oracle	DIAGNOSTIC ONLY — fit-on-test oracle T-scaling per Guo et al. 2017 [#guo]_.
reliability_curve	Bin-level calibration data wrapping sklearn.calibration.calibration_curve().
reliability_diagram_data	Structured per-bin reliability rows for serialization or plotting.