Re: [Scikit-learn-general] Randomized PCA on sparse matrix

2013-01-12 Thread Andreas Mueller
Hi. As far as I know, sparse data is not mean-centered. I don't know the poster of the answer, but I don't think the claim is true. Anyway, our documentation doesn't seem to be sufficient :-/ Best, Andy On 01/12/2013 06:19 PM, Neva Cherniavsky wrote: Hello, I'm using RandomizedPCA to get the

[Scikit-learn-general] Randomized PCA on sparse matrix

2013-01-12 Thread Neva Cherniavsky
Hello, I'm using RandomizedPCA to get the first few principal components of a large sparse matrix. However, the matrix isn't mean-centered to begin with. If I call RandomizedPCA with the full version of the matrix, the function will mean-center it before running the SVD. What happens with a spa

Re: [Scikit-learn-general] Randomized PCA

2011-11-03 Thread Olivier Grisel
2011/11/3 Mathieu Blondel : > On Thu, Nov 3, 2011 at 6:28 AM, David Warde-Farley > wrote: > >> I wonder how this compares to learning a linear tied-weights autoencoder >> with SGD and then just orthogonalizing the weight vectors (I suppose you'd >> also need to do one run with a single "neuron" in

Re: [Scikit-learn-general] Randomized PCA

2011-11-03 Thread Mathieu Blondel
On Thu, Nov 3, 2011 at 6:28 AM, David Warde-Farley wrote: > I wonder how this compares to learning a linear tied-weights autoencoder > with SGD and then just orthogonalizing the weight vectors (I suppose you'd > also need to do one run with a single "neuron" in order to orient the basis > with re

Re: [Scikit-learn-general] Randomized PCA

2011-11-02 Thread Olivier Grisel
2011/11/3 Kenneth C. Arnold : > On Wed, Nov 2, 2011 at 6:04 PM, Olivier Grisel > wrote: >> 2011/11/2 Radim Rehurek : >>> If you decide to implement the randomized PCA, I can offer some >>> observations: >>> >>> 1. oversampling does little, accuracy comes mostly from the extra power >>> iteratio

Re: [Scikit-learn-general] Randomized PCA

2011-11-02 Thread Kenneth C. Arnold
On Wed, Nov 2, 2011 at 6:04 PM, Olivier Grisel wrote: > 2011/11/2 Radim Rehurek : >> If you decide to implement the randomized PCA, I can offer some observations: >> >> 1. oversampling does little, accuracy comes mostly from the extra power >> iteration steps >> 2. no power iterations result in m

Re: [Scikit-learn-general] Randomized PCA

2011-11-02 Thread Olivier Grisel
2011/11/2 Radim Rehurek : > Hi guys, > >> Od: Olivier Grisel >> 2011/11/2 Stéfan van der Walt : >> > Hi all, >> > >> > Maybe this paper, from the current issue from SIAM Journal on >> > Scientific Computing is of some interest: >> > >> > http://epubs.siam.org/sisc/resource/1/sjoce3/v33/i5/p2580_s1

Re: [Scikit-learn-general] Randomized PCA

2011-11-02 Thread David Warde-Farley
On Wed, Nov 02, 2011 at 09:06:46PM +0100, Olivier Grisel wrote: > 2011/11/2 Stéfan van der Walt : > > Hi all, > > > > Maybe this paper, from the current issue from SIAM Journal on > > Scientific Computing is of some interest: > > > > http://epubs.siam.org/sisc/resource/1/sjoce3/v33/i5/p2580_s1?view

Re: [Scikit-learn-general] Randomized PCA

2011-11-02 Thread Olivier Grisel
2011/11/2 Stéfan van der Walt : > Hi all, > > Maybe this paper, from the current issue from SIAM Journal on > Scientific Computing is of some interest: > > http://epubs.siam.org/sisc/resource/1/sjoce3/v33/i5/p2580_s1?view=print AFAIK, Radim Rehurek in CC has already implemented this algorithm in

[Scikit-learn-general] Randomized PCA

2011-11-02 Thread Stéfan van der Walt
Hi all, Maybe this paper, from the current issue from SIAM Journal on Scientific Computing is of some interest: http://epubs.siam.org/sisc/resource/1/sjoce3/v33/i5/p2580_s1?view=print Regards Stéfan -- RSA(R) Conference