merton.backtest.calibration¶

Calibration / reliability curves for PD models.

Classes¶

CalibrationCurve

Output of calibration_curve().

Functions¶

`calibration_curve`(→ CalibrationCurve)	Reliability / calibration curve.
`calibration_plot`(→ Any)	Render a calibration plot. Returns the matplotlib Axes.

Module Contents¶

class merton.backtest.calibration.CalibrationCurve[source]¶

Output of calibration_curve().

mean_predicted: merton._typing.FloatArray[source]¶

fraction_positives: merton._typing.FloatArray[source]¶

bin_counts: merton._typing.FloatArray[source]¶

merton.backtest.calibration.calibration_curve(predictions: merton._typing.ArrayLike, defaults: merton._typing.ArrayLike, *, bins: int = 10, strategy: str = 'quantile') → CalibrationCurve[source]¶

Reliability / calibration curve.

Parameters:

bins – Number of bins.
strategy – "quantile" (default) puts equal counts in each bin; "uniform" uses equal-width bins on [0, 1].

merton.backtest.calibration.calibration_plot(predictions: merton._typing.ArrayLike, defaults: merton._typing.ArrayLike, *, bins: int = 10, strategy: str = 'quantile', ax: matplotlib.axes.Axes | None = None) → Any[source]¶: Render a calibration plot. Returns the matplotlib Axes.