Recruiting prediction model....

[volinathens]

Perhaps I misunderstand what you are saying, but I don't think that woefully naive is the correct terminology at all. The predictions is trained through various (large) training sets to establish the weighting for the regression. However, certain school could go about recruiting differently - and thus not correlate well with the mean - or a certain recruit could look for something very different than the typical recruit, and thus not correlate well. Of course the prediction can be "way off" for some people/schools and not for others, depending on how well their priorities correlate with the dominant priorities of other recruits.

I think that you and I are basically saying the exact same thing. I do think it's naive for people to pick out a couple of recruits and look at the predicted results and call the thing a farce based on those few recruits. Obviously these things don't work well when you pick out an unrepresentative sample. In UT's case, we obviously don't fit the mold very well and for people who don't understand how the model works, that can be frustrating. However, when the model is used for all 250 recruits, you are spot on when you say that it depends "on how well their priorities correlate with the dominant priorities of other recruits."

 

[BrierVol61]

Ignore this.

:blink:

 

[OptimusPrimerib]

They have still Tennessee on Tahj Boyd's list. Instant wrong

 

[UTvol1fn]

75% is pretty good, but when were the predictions made?? did they continue to change up until the player committed?

 

[PredUTor J]

Eric Gordon to Oklahoma
David Oku to Nebraska
Aaron Murray to Florida
Bryce Brown to Missouri
Chris Davenport to Hawaii
Chris Bonds to Tennessee

If even one of these came true, I would be quite impressed with their prediction model.

 

[volfanintampa]

it has giddins as going to florida but hes already verbally committed to usf

 

[volunteers4life]

hhuijhkhuihiouhiuh