weight_declustering

weight_declustering(
    coords, values, targets, *, estimator: _Base
) -> Declustering

Declustering weights from estimation weights: each sample's weight is the sum of the weights it receives when estimator estimates targets, scaled to sum to the number of samples, as in cell_declustering.

A sample alone in a sparse area informs many targets and gets a large weight; samples in a cluster share the targets around them. Targets the search leaves unestimated add nothing, so they should cover the domain and no more. Kriging can give negative weights, so a screened sample's weight can be below 0.

Parameters:

Name	Type	Description	Default
coords	(array_like, PointSet or BlockModel)	Sample locations, (n, 2) or (n, 3), or a container whose column values may name.	required
values	array_like or str		required
targets	(array_like, PointSet or BlockModel)	A dense set of points, or blocks (their centroids), covering the domain.	required
estimator	(OrdinaryKriging, SimpleKriging, UniversalKriging, BlockKriging, IndicatorKriging)	InverseDistance, NearestNeighbor or MovingAverage Its search passes and variogram, and its weights; it is not fitted or changed. NearestNeighbor gives polygon declustering at the resolution of the targets.	required

Returns:

Type	Description
Declustering	weights per sample, samples sharing a location sharing one weight, and the declustered mean; cell_size is NaN.

used in

• weight declustering