ml-modules

0.1.0 • Public • Published

Build Status GitHub GitHub code size in bytes GitHub package.json version

ml-modules

Machine learning as modules

Browser-ready machine learning algorithms as modules.

Demo

A live demo is available here.

Documentation

Documentation can be found here.

Usage

//import the modules
const modules = require("ml-modules");
//select the machine learning module you want, for example
const SVM = modules.SVM; // support vector machine

Modules

Can choose from these machine learning modules:

  • SVM
    const SVM = modules.SVM;
  • KNN
    const KNN = modules.KNN;
  • RBF
    const RBF = modules.RBF;
  • RANDF
    const RANDF = modules.RANDF;
  • LOGREG
    const LOGREG = modules.LOGREG;
  • NN
    const NN = modules.NN;


Algorithms

  • Support Vector Machine with different kernels:
    • linear
    • polynomial
    • radial-basis-function (gaussian)
  • KNN
  • Radial-basis function
  • Random Forests
  • Logistic Regression
  • Neural Net
    • multiple layers with costum definition

All algorithms are small modules. The files needed for the algorithm to work are located in his directory, except for the utility functions.

All algorithms share the basic structure.

Example

const algorithm = function() {}; // expose this function
algorithm.prototype = {
  // define the function
  train: function(data, labels) {
    //set up the environment
    //train
    //stored results
  },
  predict: function(point) {
    //returns the value predicted
  },
  predictClass: function(point) {
    //returns the class predicted
  },
  getOptions: function() {
    //returns an object to be used by the "ui" class
  },
  setOptions: function(options) {
    //set the options
  }
};
// helper functions if needed
module.exports = algorithm;

Webpack

A bundler for javascript code: you can use nodejs modules in the brower. All js files will be merged and transpiled into one bundle (index.bundle.js), generated into the ./dist folder.

To build the source code run in the command line, inside the package.json directory:

npm install

Now you have installed webpack and the project dependecies. Now you can build with:

npm run build

To be able to watch the files and automatically build on changes, just run the command:

npm run watch

Support on Beerpay

Hey dude! Help me out for a couple of 🍻!

Beerpay Beerpay

Package Sidebar

Install

npm i ml-modules

Weekly Downloads

3

Version

0.1.0

License

MIT

Unpacked Size

695 kB

Total Files

55

Last publish

Collaborators

  • davide97g