Node-Naive-Bayes

This is a Nodejs implementation of the Naive Bayes classifer. This is basically part of my final year project which also includes a real-time messaging app to show the classifier in action.

The algorithm used here is based on a book, "Programming Collective Intelligence" by Toby Segaran.

Usage

const NaiveBayes = require('node-naive-bayes');

const naiveBayes = new NaiveBayes();

naiveBayes.trainInline('the quick rabbit jumps fences', 'good');

console.log('quick rabbit: ', naiveBayes.classify('quick rabbit', 'unknown'));

You can set thresholds for a category so that the classifier does not classify an item or document wrongly when it does not have enough information

naiveBayes.setThreshold('bad', 3);

Training methods

There are two methods of training the classifier;

• Inline

  naiveBayes.trainInline('make quick money at the online casino', 'bad');

  This function accepts two parameters; first is the training text and second is the category

  Update: You can now save training data to a file so that they can be easily reused later. You do this by passing true as the third parameter and then the path to the file as the fourth.

  ...
  naiveBayes.trainInline('join bet to make excess cash in one day', 'bad', true, './my_file.txt');
  ...
  // next time just do like below, to reuse the training data
  naiveBayes.trainFromFile('./my_file.txt');

• From files

  This has the format of text:::category and are separated by new lines.

  Example is: make quick money at the online casino:::bad

  After which, you let the classifier know about the file like below

  naiveBayes.trainFromFile('path_to_file');

Contribution

Feel free to contact me or send PRs for improvements