NMD

mlquantify.metrics.NMD(prev_pred, prev_real, distances=None)

Compute the Normalized Match Distance (NMD), also known as Earth Mover’s Distance (EMD), for ordinal quantification evaluation.

Parameters:

prev_realarray-like or dict

True prevalence values for each ordered class.

prev_predarray-like or dict

Predicted prevalence values for each ordered class.

distancesarray-like of shape (n_classes-1,), optional

Distance between consecutive classes (d(y_i, y_{i+1})). If None, all distances are assumed to be 1.

Returns:

nmdfloat

Normalized Match Distance between predicted and true prevalences.