Estimate a Gaussian Mixture model using the Expectation Maximation (EM) algorithm
Fits an n-dimensional Gaussian Mixture to data using the Expectation Maximization (EM) algorithm. The algorithm is guaranteed to improve on every iteration, but it's not guaranteed to converge to a global maximum. The implementation currently uses a pretty naive initial guess. I may integrate k-nearest neighbors initilization, or may leave that to the user. I haven't decided yet. The demo uses k-nearest neighbor initialization, but the particular module is a bit unstable sometimes, so at the moment, I'm leaning toward excluding anything but rudimentary built-in initialization.
Demo at: https://rreusser.github.io/gaussian-mixture-estimator/
Not currently on npm. Pass it a list of n-dimensional coordinates (1D components must currently be wrapped in an array) and specify the number of mixture components. More documentation to come.
var estimate = require('gaussian-mixture-estimator')
estimate(
[[x1, y1, ...],
[x2, y2, ...],
[x3, y3, ...],
...
],
5
);© 2016 Ricky Reusser. MIT License.