Sampling Utilities

Algorithm overview

Sampling utilities help create subsets and train/test splits from JavaScript arrays. They are useful for small ML experiments, examples, validation flows, and browser-side demos.

JavaScript implementation

@kanaries/ml exports sampling helpers under utils.Sampling. The functions work with generic arrays, so feature rows and labels can remain in normal JavaScript data structures.

Quick start example

import { utils } from '@kanaries/ml';

const X = [[0], [1], [2], [3], [4]];
const y = [0, 0, 1, 1, 1];

const split = utils.Sampling.trainTestSplit(X, y, {
  testSize: 0.4,
  randomState: 42,
});

const sample = utils.Sampling.std([1, 2, 3, 4, 5], 3);

Detailed API reference

utils.Sampling.std<T>(arr: T[], size: number): T[]

Returns a sample without replacement. If size is greater than or equal to the array length, it returns a shallow copy.

utils.Sampling.trainTestSplit<X, Y>(
  X: X[],
  y?: Y[],
  options?: {
    testSize?: number;
    shuffle?: boolean;
    randomState?: number;
  },
): {
  XTrain: X[];
  XTest: X[];
  yTrain?: Y[];
  yTest?: Y[];
}

testSize can be a fraction less than 1 or an absolute count. shuffle defaults to true; set randomState for repeatable splits.