[pymvpa] Realtime data analysis

[Michael Hanke]

On Thu, Feb 25, 2010 at 12:53:52PM +0000, d.j.s.legge at reading.ac.uk wrote:
> I have experience in using Weka and some Matlab but am interested in
> py-mvpa as it is written in Python. I wonder if you are able to
> assist regards a question of "real-time" analysis?
> 
> That is if I have a dataset that is constantly updating is it
> possible to create a "feed" of that data into py-mvpa directly or
> via a third-party app such as mySQL?

I assume that you mainly want to do real-time classification/prediction, not
real-time training, right?

In this case you don't even need to have a dataset. PyMVPA's classifiers
can eat plain NumPy arrays in predict(). So if you get your real-time
data into such an array you are set. How your can do that obviously
depends on where your data comes from -- if it is in some SQL thingie
python-mysqldb might help -- but there are countless ways to achieve
this.


These guys

http://www.ocztechnologyforum.com/forum/archive/index.php/t-54620.html

seem to use PyMVPA for real-time analysis with this machine:

http://www.ocztechnology.com/products/ocz_peripherals/nia-neural_impulse_actuator

http://www.ocztechnologyforum.com/forum/showpost.php?p=410684&postcount=14


HTH,

Michael

-- 
GPG key:  1024D/3144BE0F Michael Hanke
http://mih.voxindeserto.de