Machine Learning Cheat Sheet (for scikit-learn)

peekaboo-vision.blogspot.de

158 points·stadeschuldt··8 comments

I had a short look at scikit-learn for a hobby classifier I'm writing (it aims to email me interesting HN articles so I don't waste much time reading things I don't like or refreshing HN all the time), and it looks fantastic. I haven't looked at the code or used it very much, but the documentation is very thorough and informative, and the library itself is very extensive.

There are some things missing, (e.g. The naive Bayes implementation is offline, which is not very useful to me), but the library is fantastic for prototyping, at least. Congratulations to the people working on it!

Thanks :) We should really add online naive Bayes and it would be really easy. But no-one got around to it yet :-/ Feel free ;)

I just might! I'll issue a pull request if I do, thanks!

For predicting structure (which takes you to 'tough luck' in the chart as it's not implemented in scikit-learn), you can use structured SVM's through either SVM-Struct (http://www.cs.cornell.edu/people/tj/svm_light/svm_struct.htm...) or JLIS (http://flake.cs.uiuc.edu/~mchang21/softwares/JLIS/indirect.h...).

Though note that both libraries are free for non-commercial use only.

Or you could use pystruct: https://github.com/amueller/pystruct But I saved that for another blog post ;)

That is BSD licensed and pure Python, has some more feature than SVMstruct but is not as optimized (yet).

I had no idea your project existed; thanks for sharing.

I did a another post on that one. http://peekaboo-vision.blogspot.de/2013/01/pystruct-more-str... The post still needs some more detail, though.

I know nothing much about machine learning (though I was digging around scikit learn earlier this evening) but this is a diagram I can appreciate :)

Reply on news.ycombinator.com