Skip to content

Latest commit

 

History

History
54 lines (37 loc) · 2.1 KB

README.md

File metadata and controls

54 lines (37 loc) · 2.1 KB

build status

credulous.js

credulous.js is a simple implementation of a Naive Bayes classifer in JavaScript. You train it with a corpus of strings, and then classify new strings.

Isn't There Already A Naive Bayes Classifier For JavaScript?

Yes, but then I wouldn't get to write this one myself!

Also, this one implements some nice trickyness. It lets you classify on multiple attirbutes, does Laplace correction to better handle infrequently found words, does calculations in log-space to avoid underflow, and has a few other improvements.

Installation

node.js

credulous.js is a CommonJS module, so it will work out of the box for node.js. Install like so:

npm install credulous

Browser

credulous.js was tested in modern browsers, and relies upon ECMAScript 5 features. That means IE 9 and above, Chrome, FireFox, and Safai.

If you want to use credulous.js in the browser, pakmanager, Ender.js, browserify or OneJS all let you use CommonJS packages in client-side code. See their documentation for details.

You can also just download a standalone version here.

How To Use It

Enough talk about usage. Get to the codes!

var Credulous = require('credulous');

var model = new Credulous();
// .train can take a single object or an array of objects
// The objects must be of the form { data: 'some data', class: 'a string representing the class name' }
model.train({ data: 'Great viagra for you!', class: 'spam' });
model.train([{ data: 'UTOSC is the best conference ever!', class: 'not spam'},
             { data: 'Some more strings that are not spam!', class: 'not spam'}]);

var result = model.classify('You have received a large inheritance, please just send me a blank money order!');
console.log('result is', result);
// 'result is spam'