Search All of the Math Forum:

Views expressed in these public forums are not endorsed by NCTM or The Math Forum.

Topic: Is there a way to calculate an average ranking from uneven lists?
Replies: 15   Last Post: Oct 30, 2013 12:18 PM

 Messages: [ Previous | Next ]
 Graham Cooper Posts: 4,495 Registered: 5/20/10
Re: Is there a way to calculate an average ranking from uneven lists?
Posted: Oct 28, 2013 3:31 AM

On Monday, October 28, 2013 12:14:36 AM UTC-7, graham...@gmail.com wrote:
>
> SITE 2
>
> BOOK 1 100%
>
> BOOK 5 80%
>
> BOOK 3 60%
>
> BOOK 11 40%
>
> BOOK 2 20%
>
> BOOK 4 0%
>
>
>
>
>
>
>
> Then to modify each books rank use:
>
>
>
>
>
> BOOKRANK = (BOOKRANK) + %*TOP - (100-%)*BOTTOM
>
> / 2
>
>
>
>
>
> If you use annealing temperatures its like this:
>
>
>
>
>
> BOOKRANK = (BOOKRANK) + TEMP*%*TOP - TEMP*(100-%)*BOTTOM
>
> / (TEMP+1)
>
>

Not quite right, with the bottom rank for each site being around 0.5
and the top ranks 2.0, they are meant to be multiplied not added.

SITE TOP BOTTOM
1 1.0 -1.2
2 1.1 -0.8
3 0.8 -0.9

That might work better with my formula!

BOOKRANK = BOOKRANK + TEMP*%*TOP + TEMP*(100-%)*BOTTOM
/ (TEMP+1)

But you'll get a good result with a simple SCALAR for each site too.

BOOKRANK = ( BOOKRANK + TEMP*WEIGHT*% ) / (TEMP + 1)

that's what I used on horse races!

That gives each RACE a CLASS which helps pick trifectas!
With books there could be a big variation in GENERAL REVIEWS (wide variance)
and specific review sites of books all in much the same score anyway...
which is where the TOP and BOTTOM more complicate spread calcs. would
improve the result..

Herc

Date Subject Author
10/27/13 Ben Bacarisse
10/28/13 Jennifer Murphy
10/30/13 Jennifer Murphy
10/27/13 James Waldby
10/28/13 Jennifer Murphy
10/28/13 Graham Cooper
10/28/13 Graham Cooper
10/28/13 Graham Cooper
10/28/13 Graham Cooper
10/28/13 Graham Cooper
10/28/13 Jennifer Murphy
10/28/13 JohnF
10/28/13 Jennifer Murphy
10/29/13 JohnF
10/30/13 Jennifer Murphy