mesmer.distrib.ProbabilityIntegralTransform.transform

mesmer.distrib.ProbabilityIntegralTransform.transform#

ProbabilityIntegralTransform.transform(data, target_name, preds_orig=None, preds_targ=None, *, threshold_proba=1e-09)#

Probability integral transform data given coefficients for the expression of a conditional distribution.

Parameters:

  • data (xr.DataTree | xr.Dataset) – Data to transform.

  • target_name (str) – name of the variable to transform

  • preds_orig (Datatree | xr.Dataset | None) – Covariants for the original distribution. Pass None, if distrib_orig does not require any.

  • preds_targ (Datatree | xr.Dataset | None) – Covariants of the target distribution. Pass None, if distrib_targ does not require any.

  • threshold_proba (float, default: 1.e-9.) – Threshold for the probability of the sample on the original distribution. The probabilities of samples outside this threshold (on both sides of the distribution) will be set to the threshold. This should avoid very unlikely values

Returns:

transf_inputs (DataTree | xr.Dataset) – Transformed data.