VotingAPI Bayesian

Introduction

I've always been fascinated by the numbers and I like playing with them.

This module has been created to fill a gap when you're using the VotingAPI with Fivestar module.
The most common use of this module will be in Views, to sort entities using values computed by a new function.

So, basically it provides a new function in the "vote results" relationship to compute a Bayesian average of each item according to the others.

Every time you submit a vote, VotingAPI compute the new value on each item.
If VotingAPI is configured to tally at cron, it will also works.

Origins

Imagine:

• 1 million users vote 4 stars on a node A
• 1 user votes 5 stars a node B

Result: node B is now better rated than node A

• #566370: bayesian averages and aggregates
• #764672: More Representative Rankings OR Weighing Star Rankings / Bayesian average

Dependencies

• Voting API

Documentation

• Computing a Bayesian Estimate of Star Rating Means
• Bayesian Ratings: Your Salvation for User-Generated Content
• Of Bayesian average and star ratings
• Using Bayesian Average to “Rank” Content Based on User Ratings
• Bayesian Average Ratings
• When can I trust an average rating on Amazon?
• Issue queue: How to pick C and M ?

Contribute back

If you want to contribute to this module and improve it in any way, feel free to post in the issue queue and submit patch. You can also contact me in private, but I prefer the issue queue.

Images by to http://xkcd.com/ under Creative Commons Attribution-NonCommercial.