LaplacePrior

class pymc_marketing.special_priors.LaplacePrior(dims=None, centered=True, **parameters)

A Laplace prior parameterized by a location and a scale parameter.

Unlike the standard Laplace distribution available through the Prior class, this distribution can optionally be centered or non-centered. A non-centered parameterization can sometimes eliminate sampling problems in hierarchical models.

Parameters:

muPrior, float, int, array_like

The location parameter of the distribution.

bPrior, float, int, array_like

The scale parameter of the distribution.

dimstuple[str, …], optional

The dimensions of the distribution, by default None.

centeredbool, optional

Whether to use the centered parameterization, by default True.

References

• A.C. Jones, Scale mixtures of unbounded continuous distributions.

• Stan Documentation, Unbounded continuous distributions.

Methods

LaplacePrior.__init__([dims, centered])
LaplacePrior.create_variable(name)	Create a variable from the prior distribution.
LaplacePrior.from_dict(data)	Create a SpecialPrior prior from a dictionary.
LaplacePrior.sample_prior([coords, name])	Sample from the prior distribution.
LaplacePrior.to_dict()	Convert the SpecialPrior to a dictionary.