Hacker News Reader

Comment by amueller | Hacker News Reader

Thanks for this great feedback :)

What algorithms do you think scikit-learn is still missing compared to R?

Largely, what I need is integration with JAGS http://mcmc-jags.sourceforge.net/ which is a niche that pymc fills. I am working on a PR to get in the other stuff I need.

I bring up R more as a point that there is more to a library than supporting lots of algorithms. R wins by that metric http://www.cran.r-project.org/web/packages/available_package...

I just want us to win on all metrics ;) at least if there is benefit for the users in it.

The comparison doesn't really make sense: R is a language (and environment), scikit-learn is a library.

It'd be really great if you guys had an implementation of Friedman's Multivariate Adaptive Regression Splines[1]. For me, this is the only R package left for which I don't have a Python alternative.

[1] http://www.salford-systems.com/doc/MARS.pdf

I'd actually be very interested in such a model. It's a major amount of work, and in an area in which I am not expert, so I know that I will not be working on it any time soon, but I agree that it would be a great addition to the scikit.

I did not know about MARS, do you have any practical benchmarking dataset / sample tasks where MARS is state of the art? What kind of regression problem do you use it for yourself? What is the typical dimension of the problems it can address efficiently (n_samples and n_features)?

Reply on news.ycombinator.com