SourceDevelop
  • Home
  • Learn A Language
  • Articles & Tutorials
  • Forums
  • Other Resources
Visual Basic HTML/CSS PHP/MySQL
PHP MySQL More Resources
index Forums Index Search Search Member List Member List Help Help

  • Search
  • Member List
  • Calendar
  • Help

Welcome Guest ( Login | Register )


Latest News: We share our revenue with our members, click here for more details.
SourceDevelop > Programming and Web Design > General Programming & Web Design Discussion > Cool formula for determining rankings

Partners:
TV & Dinner MorningStarr RustyWalrus Josechukkiri

Post Reply  Post Thread 
Threaded Mode | Linear Mode
Cool formula for determining rankings
Author Message
Tim


Bronze Junior Member
*

Group: Members
Posts: 36
Joined: Jul 2006
Status: Offline
Reputation: 1
SD$: 36
Post: #1
Cool formula for determining rankings

On the bottom of IMDb's Top 250 page, there is a notice:

Quote:
The formula for calculating the Top Rated 250 Titles gives a true Bayesian estimate:

weighted rating (WR) = (v ÷ (v+m)) × R + (m ÷ (v+m)) × C

where:
R = average for the movie (mean) = (Rating)
v = number of votes for the movie = (votes)
m = minimum votes required to be listed in the Top 250 (currently 1300)
C = the mean vote across the whole report (currently 6.7)


The first time I saw this, I thought it was genius. It solves the issue of an item having one vote of 100% and being ranked on top of an item having 100 votes with an average of 95%.

I was going to implement it into my websites, and I did a little more research on it -- basically what it does is pull the rating closer towards the mean, and the more votes it has, the farther the rating can stray. So if something has only one vote, it's rank will have to stay near the average (mean), until it gets more votes. But items with a lot of votes (a lot of general consensus) will be able to move around the ranks more. That way, you know that the item in spot #1 is the all-around best, and the item at the bottom has been agreed on being the worst.

Maybe i'm the only person here who would get excited about a sweet mathematical formula, but I thought i'd share just in case anyone else found it useful... Big Grin

edit for clarity/fix typo

07-29-2007 02:07 AM
Find all posts by this user Quote this message in a reply
Brian


Member
*

Group: Administrators
Posts: 289
Joined: Feb 2006
Status: Offline
Reputation: 0
SD$: 328
Post: #2
RE: Cool formula for determining rankings

Wow, this looks really cool. I can't stand that when websites have things where the rating is basically based off the regular rating....like you said where when it has 1 vote of the best, and it is #1...

I'll have to look into implementing this on this website for the articles and tutorials section, as right now, it simply uses the method which i really can't stand.

Thanks for the info.

Rep added


Brian
07-29-2007 05:00 AM
Find all posts by this user Quote this message in a reply
Tim


Bronze Junior Member
*

Group: Members
Posts: 36
Joined: Jul 2006
Status: Offline
Reputation: 1
SD$: 36
Post: #3
RE: Cool formula for determining rankings

Just realized that this means if an item has 0 votes, it's rating will default to the average score. Pretty nifty!

07-30-2007 04:32 AM
Find all posts by this user Quote this message in a reply
Brian


Member
*

Group: Administrators
Posts: 289
Joined: Feb 2006
Status: Offline
Reputation: 0
SD$: 328
Post: #4
RE: Cool formula for determining rankings

So does that mean that if an item has no-ranking it can be rated above an item that has a bad rating?

Also, how does one go about deciding what to set m at and how do you go about getting c?


Brian
07-30-2007 04:34 AM
Find all posts by this user Quote this message in a reply
Tim


Bronze Junior Member
*

Group: Members
Posts: 36
Joined: Jul 2006
Status: Offline
Reputation: 1
SD$: 36
Post: #5
RE: Cool formula for determining rankings

An item with no rating will get placed right in the middle--which I believe is a valid assumption. Otherwise it would fall to the very bottom with a rating of 0.

m - the higher the m is, the harder it is to get away from the mean. M can be any non-negative value.
If m is 0, there is no effect on ratings. They are not weighed.
Adjust it as you like. If it's set at 1, and most items have only 1 rating, most items will be close to the mean. If most of your items have few ratings, i'd recommend setting m somewhere around 0.25-0.5. Once you have built up votes, increase it to make it more competitive.

c - add up every single individual vote and divide by the number of ratings. myBB stores each tid/vote/ip in a table, so this should be easy. Add up all the votes and divide by the number of rows.

07-30-2007 05:31 AM
Find all posts by this user Quote this message in a reply
« Next Oldest | Next Newest »
Post Reply  Post Thread 

View a Printable Version
Send this Thread to a Friend
Subscribe to this Thread | Add Thread to Favorites

Forum Jump:

Contact Us | SourceDevelop | Return to Top | Return to Content | Lite (Archive) Mode | RSS Syndication

Powered By MyBB
Copyright © 2002-2009 MyBB Group