addressalign-toparrow-leftarrow-rightbackbellblockcalendarcameraccwcheckchevron-downchevron-leftchevron-rightchevron-small-downchevron-small-leftchevron-small-rightchevron-small-upchevron-upcircle-with-checkcircle-with-crosscircle-with-pluscontroller-playcrossdots-three-verticaleditemptyheartexporteye-with-lineeyefacebookfolderfullheartglobegmailgooglegroupshelp-with-circleimageimagesinstagramFill 1light-bulblinklocation-pinm-swarmSearchmailmessagesminusmoremuplabelShape 3 + Rectangle 1ShapeoutlookpersonJoin Group on CardStartprice-ribbonprintShapeShapeShapeShapeImported LayersImported LayersImported Layersshieldstartickettrashtriangle-downtriangle-uptwitteruserwarningyahoo

LA Machine Learning Message Board › Machine learning "development" advice, hints, and tricks

Machine learning "development" advice, hints, and tricks

Dave K.
davekale
Los Angeles, CA
Post #: 6
Would love some feedback from people about how they do machine learning "development" (i.e., programming language, packages, libraries, editors, etc.).

I for one find myself doing a lot of ad hoc coding in R (previously MATLAB), and little of research ML code I write lives up to the coding standards I was taught in school and at Google. Also, I'm working with large data sets so straightforward stuff usually takes forever to execute; I need to get a more "high-performance" set up (taking advantage of fast math libraries, parallel processing, GPU, distributed stuff, etc.).

Anyway, I'm experimenting, and I'd love to hear about other people's experiences. Here are some options I'm exploring:

* Seriously learning python and using the awesome python stuff (NumPy, SciPy, Theano, etc.).
* GPGPU programming: OpenCL / CUDA
* Hadoop + Mahout
* Clojure?

Eventually we'll compile this discussion into a "Page."
A former member
Post #: 2

RapidMiner and Weka.
A former member
Post #: 2
R, Perl and Weka
Rob Z.
user 10488294
Los Angeles, CA
Post #: 2
In addition to R, may I suggest

shogun
scikit-learn
vowpal wabbit
sofia-ml
maybe bottou's sgd package.
Dave K.
davekale
Los Angeles, CA
Post #: 7
Yeah, I've found R quite useful especially for prototyping and creating graphs/visualizations/etc., as well as for its extensive free selection of packages. However, I have an inexplicable affinity for python as a programming language, so I'm starting to gravitate toward it as my basic platform...

Here's a more specific question: how about list which programming languages you work in and estimate what percentage of time you spend in each one? Although I guess that could indicate inefficiency of the language as much as it might indicate preference... :-P
Dave K.
davekale
Los Angeles, CA
Post #: 8
LA R Users' Group is doing a meetup addressing this very topic on Tuesday, March 29, at 6:30PM on UCLA campus: http://www.meetup.com...­
AaronW
user 5064318
Tarzana, CA
Post #: 2
I tend to prototype in Numpy/Scipy.
I'm not a big fan of Rapidminer and there is a python based app thats similar called 'Orange' (http://orange.biolab....­.
Because it's Python based, It's easy create custom Py scripts where you mix and match what you want without having to use the data mining app directly.


Vince
user 34260152
Santa Monica, CA
Post #: 1
Dave, I am curious about your Matlab background. How did you start with that and made the transition to R and possibly now Python?
Powered by mvnForum

Our Sponsors

People in this
Meetup are also in:

Sign up

Meetup members, Log in

By clicking "Sign up" or "Sign up using Facebook", you confirm that you accept our Terms of Service & Privacy Policy