Declustering Operations

cell_declustering(obj, obj_region, obj_attribute, size_x, size_y, size_z, nb_off, backend=GeostatsBackend.PYTHON, rust_multithread=False)

Compute declustered weights by Cell Declustering algorithm from C. Deutsch.

Parameters:

Name	Type	Description	Default
obj	GeoRefObject	Object to decluster in x, y, z.	required
obj_region	Optional[str]	Object region or condition to apply declustering.	required
obj_attribute	str	Object attribute to apply declustering.	required
size_x	int	Cell size in x direction.	required
size_y	int	Cell size in y direction.	required
size_z	int	Cell size in z direction.	required
nb_off	int	Number of offsets.	required
backend	GeostatsBackend	Computational backend to use, for large number of points and offsets, setting to RUST will lead to faster results	PYTHON
rust_multithread	Optional[bool]	For rust backend. Focus on it for huge dataset, small size cells or large number of offsets.	False

Returns:

Type	Description
ndarray	Declustered weights $w_i$ of each data point $i$.

moving_window_declus(obj, obj_region, obj_attribute, diam_x, diam_y, diam_z, geometry)

Compute declustered weights by Moving Window algorithm.

Parameters:

Name	Type	Description	Default
obj	GeoRefObject	Object to decluster in x, y, z.	required
obj_region	Optional[str]	Object region or condition to apply declustering.	required
obj_attribute	str	Object attribute to apply declustering.	required
diam_x	int	Ellipsoid diameter in x direction.	required
diam_y	int	Ellipsoid diameter in y direction.	required
diam_z	int	Ellipsoid diameter in z direction.	required
geometry	DeclusteringGeometry	Geometry to be used for the window. BALL -> BallTree is constructed using the Mahalanobis distance. PARALLELEPIPED -> __which_in_parallelepiped	required

Returns:

Type	Description
ndarray	declustered weights $w_i$ of each data point $i$.

Note

• counts the number $n_i$ of samples inside a moving window centered on point $i$.
• weight $w_i$ is $\frac{m_v}{n_i}$ where $m_v$ is the mean of all the $n_i$.