Skip to content

curvefit.models.gaussian_mixtures.GaussianMixtures

Fit a linear combination of some Gaussian distributions

The GaussianMixtures model allows you to fit weights for a linear combination of Gaussian distributions using least squares. It is typically used on top of a CoreModel.

Arguments

  • stride (int): the distance between the Gaussian atoms; how far to space them out over the independent variable
  • size (int): number of Gaussian atoms to include, replicated over different locations determined by stride
  • params (np.array): parameters to build the Gaussian distributions; must be parameters from a Gaussian family function in curvefit.core.functions

Methods

compute_design_matrix

Calculates the design matrix from the Gaussian atoms that will be used in order to get their optimal linear combination.

  • t (np.array): how far in time should the predictions that will go into the design matrix extend

See additional methods in CoreModel.