Re: [Scikit-learn-general] Bugfix for SpectralClustering, maybe same bug somewhere else?

On Mon, Aug 27, 2012 at 01:55:00PM +0200, Andreas Mueller wrote: > I got the problem now, it should have been > laplacian.flat[::n_nodes + 1]. Aaarrrgg. My fault -- Live Security Virtual Conference Exclusive live even

Re: [Scikit-learn-general] wiserf vs. sklearn RF

2012/8/27 Peter Prettenhofer : > Hi folks, > > I just stumbled upon this benchmark comparing wiserf, R randomForest, > sklearn (0.12), and weka: > > http://wise.io/wiserf.html > > Lets put it that way: I think we can/should do better ;-) They don't give the number of estimators nor the predictive

Re: [Scikit-learn-general] wiserf vs. sklearn RF

I wonder what their core tree algorithms are and how fast they are. It seems they're achieving those speeds by either A) Significantly optimizing the tree learner or B) Using a smart cutoff on the size of the forest based Tommy Guy On Mon, Aug 27, 2012 at 10:14 AM, Jaques Grobler wrote: > cool l

Re: [Scikit-learn-general] wiserf vs. sklearn RF

cool little interactive benchmark. Wonder what they do RF wise to train so damn quickly.. clever optimaztion or something? Agree regarding the other datasets 2012/8/27 Gilles Louppe > (However, I'd be curious to see the same benchmark on other datasets, > with a larger number of features...) >

Re: [Scikit-learn-general] Bugfix for SpectralClustering, maybe same bug somewhere else?

I got the problem now, it should have been laplacian.flat[::n_nodes + 1]. On 08/27/2012 01:48 PM, Andreas Mueller wrote: Hi everybody. Today I fixed a pretty bad bug in SpectralClustering, that cause

[Scikit-learn-general] Bugfix for SpectralClustering, maybe same bug somewhere else?

Hi everybody. Today I fixed a pretty bad bug in SpectralClustering, that caused one example to be in it's own cluster and lead to a incorrect graph laplacian. Olivier suggested I write to the list to p

Re: [Scikit-learn-general] Self-organizing map, implemented with Numba

2012/8/24 Alex Wiltschko : > Hello all, > > Long time user of scikits-learn, first time posting here. > I implemented a self-organizing (aka Kohonen) map with Numba. It's fast. I'd > like to share it with the scientific python community. Can I contribute it > to scikits-learn? numba (because of ll

Re: [Scikit-learn-general] Self-organizing map, implemented with Numba

Hey Alex. This sounds interesting - I'm not familiar with the technique. We already have Manifold learning which does a non-linear dimensionality reduction as well. I don't know how it compares to the Kohonen maps but it seems worth looking into. Here's the Manifold learning documentation http://s

[Scikit-learn-general] Self-organizing map, implemented with Numba

Hello all, Long time user of scikits-learn, first time posting here. I implemented a self-organizing (aka Kohonen) map with Numba. It's fast. I'd like to share it with the scientific python community. Can I contribute it to scikits-learn? Here's a notebook with a run-down of how it works: http:

Re: [Scikit-learn-general] Getting confidence of a svc predict_prob

2012/8/26 David Montgomery : > I was going down the route of bootstrappinga lot... and distributing > across multiple cores and machines is not an issue for deriving a mean an > variance. What I am confused about is the effect, from my understanding, is > the implicit 5-Kfold used to genera

Re: [Scikit-learn-general] Issue with StratifiedShuffleSplit

Yes, that branch fixes the issue. Thanks. Dan. On 26 August 2012 11:42, Olivier Grisel wrote: > Can you check whether this branch fixes the issue? > > https://github.com/scikit-learn/scikit-learn/pull/1060 > > Would be good to include your failed examples as a non regression > tests for this P

Re: [Scikit-learn-general] wiserf vs. sklearn RF

(However, I'd be curious to see the same benchmark on other datasets, with a larger number of features...) On 27 August 2012 09:13, Gilles Louppe wrote: > Hi Peter, > > At least we are better than Weka! More seriously, this indeed shows > that there is still a lot of work to do... :( > > Gilles >

Re: [Scikit-learn-general] wiserf vs. sklearn RF

Hi Peter, At least we are better than Weka! More seriously, this indeed shows that there is still a lot of work to do... :( Gilles On 27 August 2012 09:06, Peter Prettenhofer wrote: > Hi folks, > > I just stumbled upon this benchmark comparing wiserf, R randomForest, > sklearn (0.12), and weka:

[Scikit-learn-general] wiserf vs. sklearn RF

Hi folks, I just stumbled upon this benchmark comparing wiserf, R randomForest, sklearn (0.12), and weka: http://wise.io/wiserf.html Lets put it that way: I think we can/should do better ;-) best, Peter -- Peter Prettenhofer --