Distance Metrics

This is an implementation of commonly used distance metrics in pure javascript (with typescript) as noted in this stackexchange answer. The initial motivation was to be able to compute distance metrics between histograms.

Installation

$ npm install --save distance-metrics

Example

const { euclideanDistance } = require('distance-metrics');
// Create two vectors for the points (-7, -4), (17, 6.5)
const vector1 = [-7, -4]
const vector2 = [17, 6.5]
console.log( euclideanDistance(vector1, vector2) )

Output

26.196373794859472

Distance Metrics Functions

All currently implemented distance metrics use parameters that are just arrays.

Minkowski Family

euclideanDistance(vector1, vector2)

manhattanDistance(vector1, vector2)

fractionalDistance(vector1, vector2, exponent)

minkowskiDistance(vector1, vector2, exponent)

Other Distance Functions

hellingerDistance(vector1, vector2)

chebyshevDistance(vector1, vector2)

matchDistance(vector1, vector2)

Native Distance Metrics Functions

I've started some experimental work where I'm going to implement native versions of the javascript functions. Currently I have implemented the following functions in Rust. manhattanDistanceNative(vector1, vector2)

Notes:

The manhattan, euclidean, and fractional distance functions are all special cases of the minkowskiDistance function. If you are only using function calls to minkowskiDistance(...) that are equivalent your code will be slightly faster if you call the underlying special cases directly.

manhattanDistance(vector1, vector2) is equivalent to minkowskiDistance(vector1, vector2, 1) euclideanDistance(vector1, vector2) is equivalent to minkowskiDistance(vector1, vector2, 2) fractionalDistance(vector1, vector2, <any exponent less than 1>) is equivalent to minkowskiDistance(vector1, vector2, <any exponent less than 1>)

Dev Quickstart

$ npm install
$ npm run compile