Tell me more ×
Cross Validated is a question and answer site for statisticians, data analysts, data miners and data visualization experts. It's 100% free, no registration required.
up vote 5 down vote favorite
2

There seem to be several options available for working with Gaussian Mixture Models (GMMs) in Python. At first glance there are at least:

... and perhaps others. They all seem to provide the most basic needs for GMMs, including creating and sampling, parameter estimation, clustering, etc.

What's the difference between them, and how should one go about determining which is best suited for a particular need?

Ref: http://www.scipy.org/Topical_Software

share|improve this question
You could try profiling a really simple version of image compression using GMMs. Given an image use a GMM to assign the pixels different probabilities and then recreate the image by using the probabilities as indices for the particular Gaussian that a particular pixel was most likely to have come from. – cpcloud Feb 27 '12 at 3:18
@cpcloud -- Do you mean: set up a simplified experiment to execute in each of these packages, as a point of comparison? Well, okay, but that's no small amount of effort. I am hoping for some input from people who have used these packages. – Aman Feb 27 '12 at 3:47
1  
Scikit-learn is a popular machine learning library that also has some GMM support. I am not sure it fits your needs but it has the benefit of having other learning algorithms and framework (e.g. cross-validation, model composition). – Bitwise Apr 6 at 2:56

1 Answer

up vote 1 down vote

I do not know how to determine in general which one is best, but if you know your application setting well enough, you can simulate data and try the packages on these simulation. Success metrics could be the time that the estimation take and the quality of recovery of your simulated ground truth.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.