On Oct 7, 2012, at 22:36 , Gael Varoquaux <[email protected]> wrote:
> Indeed, a non-negative least square estimator in the scikit-learn linear > models would be great. As you mention, the logic for nnls minimization is > already in scipy and we could reuse it. The word basically boils down to > writing the estimator object, the tests, examples and documentation. There is also the `_nls_subproblem` in `decomposition/nmf.py`. If I remember correctly from the last time I looked into this, the `scipy.optimize.nnls` is an order of magnitude faster under the preconditions that y is 1-dimensional and that we don't have a sensible initial guess. `_nls_subproblem` supports 2d y, initial values and returns the gradient and the number of iterations, which are used in the NMF implementation. It's on my list to understand how the two optimization algorithms actually work (it's a bit out of my comfort zone at the moment). A way to speedup `_nls_subproblem` should visibly speed up NMF too. Vlad > > We'd gladly accept a pull request implementing these. > > Thanks! > > ------------------------------------------------------------------------------ > Don't let slow site performance ruin your business. Deploy New Relic APM > Deploy New Relic app performance management and know exactly > what is happening inside your Ruby, Python, PHP, Java, and .NET app > Try New Relic at no cost today and get our sweet Data Nerd shirt too! > http://p.sf.net/sfu/newrelic-dev2dev > _______________________________________________ > Scikit-learn-general mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/scikit-learn-general ------------------ Vlad N. http://vene.ro ------------------------------------------------------------------------------ Don't let slow site performance ruin your business. Deploy New Relic APM Deploy New Relic app performance management and know exactly what is happening inside your Ruby, Python, PHP, Java, and .NET app Try New Relic at no cost today and get our sweet Data Nerd shirt too! http://p.sf.net/sfu/newrelic-dev2dev _______________________________________________ Scikit-learn-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
