Observation

Negative Binomial

class NegativeBinomialObservation(concentration_prior, concentration_suffix='_concentration', parameter_name='negbinom_rv', eps=1e-10)

Bases: RandomVariable

Negative Binomial observation

sample(mu, obs=None, name=None, **kwargs)

Sample from the negative binomial distribution

Parameters:

• mu (ArrayLike) – Mean parameter of the negative binomial distribution.

• obs (ArrayLike, optional) – Observed data, by default None.

• name (str, optional) – Name of the random variable if other than that defined during construction, by default None (self.parameter_name).

• **kwargs (dict, optional) – Additional keyword arguments passed through to internal sample calls, should there be any.

Return type:

tuple

static validate(concentration_prior)

Check that the concentration prior is actually a nums.Number

Parameters:

concentration_prior (any) – Numpyro distribution from which to sample the positive concentration parameter of the negative binomial. Expected dist.Distribution or numbers.nums

Return type:

None

Poisson

class PoissonObservation(parameter_name='poisson_rv', eps=1e-08)

Bases: RandomVariable

Poisson observation process

sample(mu, obs=None, name=None, **kwargs)

Sample from the Poisson process

Parameters:

• mu (ArrayLike) – Rate parameter of the Poisson distribution.

• obs (ArrayLike | None, optional) – Observed data. Defaults to None.

• name (str | None, optional) – Name of the random variable. Defaults to None.

• **kwargs (dict, optional) – Additional keyword arguments passed through to internal sample calls, should there be any.

Return type:

tuple

static validate()

Validation of kwargs to be implemented in subclasses.