learn_spatial_weights

learn_spatial_weights(side=12, rho=0.6, k=4, permutations=199, seed=0)

See how the choice of spatial weights changes what “neighbors” means.

Simulates one field with dependence rho under queen contiguity (the true graph), then re-tests the same field under queen, rook and k-nearest-neighbor weights. All three detect the clustering, but the statistic shifts with the graph — the substantive conclusion should not hinge on one W, which is why :func:geometrics.analyze_spatial_model_by_weights exists.

Parameters

Name	Type	Description	Default
side	int	Lattice side length (n = side²).	12
rho	float	Planted spatial dependence under the queen graph, |ρ| < 1.	0.6
k	int	Neighbors for the k-nearest-neighbor variant.	4
permutations	int	Conditional permutations behind each pseudo p-value.	199
seed	int	Random seed.	0

Returns

Name	Type	Description
	SandboxResult	df (one row per weights choice), fig, summary, topic and the simulated field in data.

Examples

import geometrics as gm

res = gm.learn_spatial_weights(rho=0.6, k=8)
res.df