pysteps.postprocessing.ensemblestats.banddepth

pysteps.postprocessing.ensemblestats.banddepth(X, thr=None, norm=False)

Compute the modified band depth (Lopez-Pintado and Romo, 2009) for a k-member ensemble data set.

Implementation of the exact fast algorithm for computing the modified band depth as described in Sun et al (2012).

Parameters:

• X (array_like) – Array of shape (k, m, …) representing an ensemble of k members (i.e., samples) with shape (m, …).

• thr (float) – Optional threshold for excluding pixels that have no samples equal or above the thr value.

Returns:

out – Array of shape k containing the (normalized) band depth values for each ensemble member.

Return type:

array_like

References

Lopez-Pintado, Sara, and Juan Romo. 2009. “On the Concept of Depth for Functional Data.” Journal of the American Statistical Association 104 (486): 718–34. https://doi.org/10.1198/jasa.2009.0108.

Sun, Ying, Marc G. Genton, and Douglas W. Nychka. 2012. “Exact Fast Computation of Band Depth for Large Functional Datasets: How Quickly Can One Million Curves Be Ranked?” Stat 1 (1): 68–74. https://doi.org/10.1002/sta4.8.