On Tue, Nov 29, 2011 at 5:24 PM, Olivier Grisel
<[email protected]> wrote:
> That makes sense. Fortunately we don't have an API to compute the
> expected variance of a prediction :)

Slightly off-topic, but this is exactly what's necessary to use
existing regression algorithms for Bayesian optimization, even
internally for hyper-parameter optimization.  So not crazy to think
about adding such an API.

Also, I think it at least a few cases it's more efficient to ask for
both the mean and variance of a conditional prediction at the same
time, rather than asking first for the mean, and then second for the
variance. They can often share a lot of computation (GP is one
example).  What would be the most scikits-friendly way of dealing with
that?

- James

------------------------------------------------------------------------------
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