forecaster.HMCForecaster

Fit a forecasting model with MCMC (NUTS by default).

Usage

Source

forecaster.HMCForecaster(
    rng_key,
    model,
    data,
    covariates,
    *,
    kernel=None,
    kernel_kwargs=None,
    num_warmup=1000,
    num_samples=1000,
    num_chains=1,
    chain_method="sequential",
    progress_bar=False
)

Parameters

rng_key: Array

PRNG key for inference.

model: ForecastModel

The forecasting model to fit (OOP instance or functional model).

data: Array

In-sample data with time at axis -2.

covariates: Array

Covariates with time at axis -2 and the same duration as data.

kernel: KernelLike = None

Kernel specification resolved by ~numpyro_forecast.functional.mcmc.resolve_kernel(): None (NUTS), an MCMCKernel instance, or an MCMCKernel subclass.

kernel_kwargs: Mapping[str, Any] | None = None

Extra keyword arguments for the kernel constructor (only with None or a kernel class).

num_warmup: int = 1000

Number of warmup steps.

num_samples: int = 1000

Number of posterior samples.

num_chains: int = 1

Number of MCMC chains.

chain_method: str = "sequential"

NumPyro chain method ("sequential"/"parallel"/"vectorized").

progress_bar: bool = False

Whether to display the MCMC progress bar.