[Dev-flock] bayesian weighted rating system

[Clay Ward]

Is that all there is to Bayesian rating?  I guess we originally thought that
there was a trick with an infinite loop as the equations end up referencing
themselves.  But I think we've grown comfortable with the idea that things
can be calculated based on their current state and not need to refer to
themselves.

I've attached an adaption of the Bayesian equations that should help us
figure out our evaluation order.  Please modify as needed.

-clay

On Tue, Apr 1, 2008 at 10:18 PM, Lucy Mendel <lucy at thoughtandmemory.org>
wrote:

> (holy bat cave, batman)
>    http://www.thebroth.com/blog/118/bayesian-rating
> from
>   http://reddit.com/goto?id=6e8xq
> incidentally, the current trend is on data mining and datasets
>   http://www.datawrangling.com/some-datasets-available-on-the-web.html
> _______________________________________________
> Dev-flock mailing list
> Dev-flock at mit.edu
> http://mailman.mit.edu/mailman/listinfo/dev-flock
>



-- 
Clay Ward
President and Founder
Thought And Memory, Inc
http://www.ThoughtAndMemory.org

"Buy It Like You Mean It"
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.mit.edu/pipermail/dev-flock/attachments/20080402/a3fbf01b/attachment.htm
-------------- next part --------------
A non-text attachment was scrubbed...
Name: EvaluationOrder.rtf
Type: application/rtf
Size: 1651 bytes
Desc: not available
Url : http://mailman.mit.edu/pipermail/dev-flock/attachments/20080402/a3fbf01b/attachment.rtf