bayes_naive_jdp

BayesNaiveJdp

This is a very simple Naive Bayesian classifier, build just for fun.

I decided to use it as an example while learning to package ruby code.

The algorithm used here is not original, but an adaptation from Burak Kanber's Machine Learning in Javascript series.

Installation

Add this line to your application's Gemfile:

gem 'bayes_naive_jdp'

And then execute:

$ bundle

Or install it yourself as:

$ gem install bayes_naive_jdp

Usage

#	create an instance of the Classifier
classifier = BayesNaiveJdp::Classifier.new

#	Train the Classifier using many samples from each valid category
#	documents are strings, categories are strings, symbols, or integers
classifier.train(document_a_1, category_a)
classifier.train(document_a_2, category_a)
classifier.train(document_a_3, category_a)
..
classifier.train(document_b_1, category_b)
classifier.train(document_b_2, category_b)
classifier.train(document_b_3, category_b)
..

#	use the classifier to categorize an unknown document
results = classifier.classify(mystery_document)
category = results[:winner][:classification]
confidence = results[:winner][:confidence]

#	also, results[:all_scores] is a hash 
#	with categories as indices and confidence levels as values