GeoDaLib

Appearance

geodalib / core/src / getKernelKnnWeightsFromGeomCollection

Function: getKernelKnnWeightsFromGeomCollection() ​

getKernelKnnWeightsFromGeomCollection(__namedParameters): Promise<number[][]>

Defined in: core/src/weights/kernel-knn-weights.ts:93

Calculates k-nearest neighbor kernel weights for a given set of geometries.

Example ​

ts
import { getKernelKnnWeightsFromGeomCollection } from '@geoda/core';

const geometries = [
  { type: 'Feature', geometry: { type: 'Point', coordinates: [0, 0] } },
  { type: 'Feature', geometry: { type: 'Point', coordinates: [1, 0] } },
  { type: 'Feature', geometry: { type: 'Point', coordinates: [0, 1] } },
];

const weights = await getKernelKnnWeightsFromGeomCollection({
  k: 2,
  kernel: 'gaussian',
  geomCollection: geometries,
});

console.log(weights);

Parameters ​

__namedParameters ​

adaptiveBandwidth? ​

boolean = true

Whether to use each observation's k-th nearest neighbor distance as its bandwidth (true) or a single global maximum distance (false).

geomCollection ​

GeometryCollection

The geometry collection to calculate the weights for.

isInverse? ​

boolean = false

Whether to apply inverse distance weighting before the kernel.

isMile? ​

boolean = false

The unit of distance (mile or km).

k ​

number

The number of nearest neighbors.

kernel ​

string

The kernel function (triangular, uniform, epanechnikov, quartic, gaussian).

power? ​

number = 1.0

The power (or exponent) used by the inverse distance weighting (1 / distance^power). Only applied when isInverse is true.

useKernelDiagonals? ​

boolean = false

Whether the diagonal (self) weight is kernel(0.0) instead of 1.0.

Returns ​

Promise<number[][]>

  • The kernel weights as rows of interleaved [index, weight] pairs.