Censored#

class pymc_extras.prior.Censored(distribution: InstanceOf[Prior], lower: float | InstanceOf[pt.TensorVariable] = -inf, upper: float | InstanceOf[pt.TensorVariable] = inf)[source]#

Create censored random variable.

Examples

Create a censored Normal distribution:

from pymc_extras.prior import Prior, Censored

normal = Prior("Normal")
censored_normal = Censored(normal, lower=0)

Create hierarchical censored Normal distribution:

from pymc_extras.prior import Prior, Censored

normal = Prior(
    "Normal",
    mu=Prior("Normal"),
    sigma=Prior("HalfNormal"),
    dims="channel",
)
censored_normal = Censored(normal, lower=0)

coords = {"channel": range(3)}
samples = censored_normal.sample_prior(coords=coords)

__init__(*args: Any, **kwargs: Any) → None#

Methods

`__init__`(args, *kwargs)
`create_likelihood_variable`(name, mu, observed)	Create observed censored variable.
`create_variable`(name[, xdist])	Create censored random variable.
`from_dict`(data)	Create a censored distribution from a dictionary.
`sample_prior`([coords, name, xdist])	Sample the prior distribution for the variable.
`to_dict`()	Convert the censored distribution to a dictionary.
`to_graph`()	Generate a graph of the variables.

Attributes

`dims`	The dims from the distribution to censor.
`lower`
`upper`
`distribution`