> Cool, good to know!
>
> On the name though --- "eval_MSE" is a nonstandard term for "variance"
> no? MSE usually refers to a loss criterion, for comparing predictions
> with targets.
>
MSE stands for "mean squared error". The GP predictor indeed ensures 
minimum prediction variance (aka the mean squared error of the best 
linear unbiased predictor). See the book by Santner et al., 2003 "Design 
and analysis of computer experiments" for the fundamental theorem of 
prediction that establishes a strict equivalence between the GP 
predictor viewed as the distribution of the prediction conditional onto 
the observations (as exposed in Rasmussen & Williams' book) and the 
"best linear unbiased predictors" (commonly used in geostatistics and 
know as the Kriging predictors). Santner's book can be partly previewed 
on Google books. I used it in my thesis to introduce Gaussian process 
predictors. I'll let you know as soon as it's available for download 
(before x-mas).

Cheers,
Vincent

------------------------------------------------------------------------------
All the data continuously generated in your IT infrastructure 
contains a definitive record of customers, application performance, 
security threats, fraudulent activity, and more. Splunk takes this 
data and makes sense of it. IT sense. And common sense.
http://p.sf.net/sfu/splunk-novd2d
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general

Reply via email to