calibration

Define calibration helpers for ss.Calibration()

Functions

Name	Description
euclidean	Euclidean distance between expected and predictec/simulated data
normalized_median_absolute_error	The normalized median absolute error. A highly robust goodness-of-fit metric.
weighted_euclidean	Weighted Euclidean distance between expected and predictec/simulated data. Also called

calibration.euclidean(expected, predicted)

Euclidean distance between expected and predictec/simulated data

Name	Type	Description	Default
expected	pd.DataFrame	dataframe with column “x”, the quantity or metric of interest, from the reference dataset.	required
predicted	pd.DataFrame	dataframe with column “x”, the quantity or metric of interest, from simulated dataset.	required

Name	Type	Description
nll	float	negative Euclidean distance between expected and predicted values.

calibration.normalized_median_absolute_error(expected, predicted)

The normalized median absolute error. A highly robust goodness-of-fit metric.

Name	Type	Description	Default
expected	pd.DataFrame	dataframe with column “x”, the quantity or metric of interest, from the reference dataset.	required
predicted	pd.DataFrame	dataframe with column “x”, the quantity or metric of interest, from simulated dataset.	required

Name	Type	Description
ngof	float	negative goodness of fit – the Calibration class tries to maximise the score of the objective function.

calibration.weighted_euclidean(expected, predicted)

Weighted Euclidean distance between expected and predictec/simulated data. Also called weighted_squares in calibra ll calculators.

Name	Type	Description	Default
expected	pd.DataFrame	dataframe with column “x”, the quantity or metric of interest, from the reference dataset.	required
predicted	pd.DataFrame	dataframe with column “x”, the quantity or metric of interest, from simulated dataset.	required

Name	Type	Description
nll	float	negative weighted Euclidean distance between expected and predicted values.