esda.prominence¶

esda.prominence(X, connectivity, return_all=False, gdf=None, verbose=False, middle='mean', progressbar=False)[source]¶

Return the prominence of peaks in input, given a connectivity matrix.

Parameters:¶

X : numpy.ndarray¶

an array of shape N,p containing data to use for computing prominence. When p > 1, X will be converted to an “elevation” using to_elevation.

connectivity : scipy.sparse matrix¶

a sparse matrix encoding the connectivity graph pertaining to rows of X. If coordinates are provided, they must be (N,2), and the delaunay triangulation will be computed.

return_class : bool (default: False)

whether or not to return additional information about the result, such as the set of dominating peaks or the set of classifications for each observation.

verbose : bool (default: None)¶

whether or not to print extra information about the progress of the algorithm.

middle : string or callable (default: "mean")¶

how to compute the center of mass from X, when the dimension of X > 2.

Returns:¶

• the prominence of each observation in X, possibly along with the

• set of saddle points, peaks, and/or dominating peak tree.

Notes

An observation has 0 prominence when it is a saddle point. An observation has positive prominence when it is a peak, and this is computed as the elevation of the peak minus the elevation of the saddle point.

Observations have “NA” prominence when they are neither a saddle point nor a peak.