neurotools.stats.mixtures module

Functions relating to distributions. Most of this should be available in numpy, scipy, and scikits.

neurotools.stats.mixtures.two_class_poisson_mixture_model(counts)[source]

Estimates a Poisson mixture model with two distributions Originally written as a toy example

Parameters:

counts (np.array) – Array of count observations. We presume that counts arise from a misture of two Poisson distributions with different means.

Returns:

  • classes (np.array) – assigned classes of points

  • mu0 (float) – mean of class 0

  • mu1 (float) – mean of class 1

  • pr0 (np.array) – probability that each count obsrevation belongs to class 0

  • pr1 (np.array) – probability that each count obsrevation belongs to class 1