Unsupervised-KNN-JS

Node.JS package for computing the k nearest neighbors to an input vector using distance calculations. Computations are implemented in Rust for high perfromance and parallelism.

Table of Contents

• Features
• Install
• Import
• Use Example
• Usage
  • Parameters
  • Return
• Distance Comparisons
• Future Features
• Changes

Features

• Parallelized distance computations
• Fast native system processing
• 14 popular distance functions
• Out of the box support on Linux, OSX, and Windows
• Support for Node 8, 10, 12, and 13

Install

$ npm i unsupervised-knn-js

Import

const { knn } = require('unsupervised-knn-js')

Example

> const { knn } = require('unsupervised-knn-js')
 
> const neighbors = [
  { label: 'some name', vector: [1, 2, 4, 5] },
  { label: 'name 2', vector: [14, 4, 13, 2] },
  { label: 'another name', vector: [4, 4, 4, 5] },
]
> const target = [1, 2, 3, 4]
> const algo = 'euclidean'
> const k = 2
 
> knn(algo, k, neighbors, target)
[
  { label: 'some name', distance: 1.4142135623730951 },
  { label: 'another name', distance: 3.872983346207417 }
]
> 

Usage

Parameters

The knn function takes 4 parameters:

1. Algorithm String
   • This is the algorithm which computes distances between the target and all neighbors
   • The current algorithms natively supported are:

       'euclidean'  // L2 Norm Difference
       'cosine'     // Cosine Distance
       'mae'        // Mean-Absolute-Error
       'mse'        // Mean-Squared-Error 
       'manhattan'  // Sum of Absolute Differences
       'ssd'        // Sum of Squared Differences
       'canberra'   // Weighted Manhatten Distance
       'hamming'    // Sum of Binary Differences
       'L3'         // L3 Norm Difference
       'L4'         // L4 Norm Difference
       'L5'         // L5 Norm Difference
       'L10'        // L10 Norm Difference
       'chebyshev'  // L-Infinite Norm Difference
       'pearson'    // Pearson Correlation Distance

2. K-Value
   • The amount of closest neighbors to the target point to return
   • So if k = 2, the 2 closests neighbors to the target vector will be returned.
3. Neighbors
   • This is an array of objects where each object represents a neighbor or point
   • Each object should have a label and vector field as such:

     {
       label: 'name or id',
       vector: [1, 3, 4.5, -4]
     }

   • The following is a valid array of neighbors:

     const neighbors = [
       { label: 'some name', vector: [1, 2, 4, 5] },
       { label: 'name 2', vector: [14, 4, 13, 2] },
       { label: 'another name', vector: [4, 4, 4, 5] },
     ]

4. Target
   • This is the vector for which to find the closest or most similar points to
   • This should be an array of numbers

Return

The function returns an array of objects representing the closest points to the target.

Each object has a label field for identification and a distance field which represents it's difference from the target.

[
  { label: 'some name', distance: 1.4142135623730951 },
  { label: 'another name', distance: 3.872983346207417 }
]

This list is ordered in ascending order based on the distance field in each object.

Distance Comparisons

Here is an example of the same data run against different distance functions

> const { knn } = require('unsupervised-knn-js')
> const neighbors = [
  { label: 'some name', vector: [1, 2, 4, 5] },
  { label: 'another name', vector: [4, 4, 4, 5] },
  { label: 'name 3', vector: [14, 4, 13, 2] },
]
> const target = [1, 2, 3, 4]
 
> // Euclidean
> knn('euclidean', 3, neighbors, target)
[
  { label: 'some name', distance: 1.4142135623730951 },
  { label: 'another name', distance: 3.872983346207417 },
  { label: 'name 3', distance: 16.64331697709324 }
]
 
> // Cosine
> knn('cosine', 3, neighbors, target)
[
  { label: 'some name', distance: 0.003993481192393733 },
  { label: 'another name', distance: 0.059777545024485734 },
  { label: 'name 3', distance: 0.35796589482505503 }
]
 
> // Mean-Absolute-Error 
> knn('mae', 3, neighbors, target)
[
  { label: 'some name', distance: 0.5 },
  { label: 'another name', distance: 1.75 },
  { label: 'name 2', distance: 6.75 }
]
 
> // Mean-Squared-Error
> knn('mse', 3, neighbors, target)
[
  { label: 'some name', distance: 0.5 },
  { label: 'another name', distance: 3.75 },
  { label: 'name 3', distance: 69.25 }
]
 
> // Manhattan
> knn('manhattan', 3, neighbors, target)
[
  { label: 'some name', distance: 2 },
  { label: 'another name', distance: 7 },
  { label: 'name 3', distance: 27 }
]
 
> // Sum of Squared Differences
> knn('ssd', 3, neighbors, target)
[
  { label: 'some name', distance: 2 },
  { label: 'another name', distance: 15 },
  { label: 'name 2', distance: 277 }
]
 
> // Canberra
> knn('canberra', 3, neighbors, target)
[
  { label: 'some name', distance: 0.25396825396825395 },
  { label: 'another name', distance: 1.1873015873015873 },
  { label: 'name 3', distance: 2.158333333333333 }
]
 
> // Hamming
> knn('hamming', 3, neighbors, target)
[
  { label: 'some name', distance: 2 },
  { label: 'another name', distance: 4 },
  { label: 'name 3', distance: 4 }
]
 
> // L3 Norm Difference
> knn('L3', 3, neighbors, target)
[
  { label: 'some name', distance: 1.2599210498948732 },
  { label: 'another name', distance: 3.332221851645953 },
  { label: 'name 3', distance: 14.756054203376182 }
]
 
> // L4 Norm Difference
> knn('L4', 3, neighbors, target)
[
  { label: 'some name', distance: 1.189207115002721 },
  { label: 'another name', distance: 3.1543421455299043 },
  { label: 'name 3', distance: 14.016098305349052 }
]
 
> // L5 Norm Difference
> knn('L5', 3, neighbors, target)
[
  { label: 'some name', distance: 1.148698354997035 },
  { label: 'another name', distance: 3.0796116495812957 },
  { label: 'name 3', distance: 13.635466232760923 }
]
 
> // L10 Norm Difference
> knn('L10', 3, neighbors, target)
[
  { label: 'some name', distance: 1.0717734625362931 },
  { label: 'another name', distance: 3.0051723058500506 },
  { label: 'name 2', distance: 13.091355843137347 }
]
 
> // Chebyshev
> knn('chebyshev', 3, neighbors, target)
[
  { label: 'some name', distance: 1 },
  { label: 'another name', distance: 3 },
  { label: 'name 3', distance: 13 }
]
 
> // Pearson Correlation Distance
> knn('pearson', 3, neighbors, target)
[
  { label: 'some name', distance: 0.010050506338833642 },
  { label: 'another name', distance: 0.2254033307585166 },
  { label: 'name 3', distance: 1.5685785754425927 }
]

Future Features

• Even more native distance functions
• Potential implemention of custom distance functions passed in by the user

Ideas and suggestions are welcome!

Changes

For changes please see the Changelog