2013/1/4 Andreas Mueller <[email protected]>:
> On 01/04/2013 12:41 AM, Vlad Niculae wrote:
>> Maybe my mind is not in its right place but how is that different from
>> using the PCA transformer?
>>
> Not sure about my mind either but Lars talked about the SVD of X, not
> the covariance matrix.

Indeed. In practical terms: when PCA subtracts the columnwise mean of
X both in fit and in transform, it destroys the sparsity, so it's no
good for sets of documents. See also
http://metaoptimize.com/qa/questions/3340/difference-between-lsi-and-pca

-- 
Lars Buitinck
Scientific programmer, ILPS
University of Amsterdam

------------------------------------------------------------------------------
Master HTML5, CSS3, ASP.NET, MVC, AJAX, Knockout.js, Web API and
much more. Get web development skills now with LearnDevNow -
350+ hours of step-by-step video tutorials by Microsoft MVPs and experts.
SALE $99.99 this month only -- learn more at:
http://p.sf.net/sfu/learnmore_122812
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general

Reply via email to