srvar.results

class srvar.results.FEVDResult(variables, shocks, horizons, draws, mean, quantiles, identification, metadata=<factory>)[source]

Bases: object

Forecast error variance decomposition (FEVD) draws and summaries.

draws: ndarray
horizons: list[int]
identification: str
mean: ndarray
metadata: dict[str, Any]
quantiles: dict[float, ndarray]
shocks: list[str]
variables: list[str]
class srvar.results.FitResult(dataset, model, prior, sampler, posterior, latent_dataset=None, latent_draws=None, beta_draws=None, sigma_draws=None, q_draws=None, h_draws=None, h0_draws=None, sigma_eta2_draws=None, sv_gamma0_draws=None, sv_phi_draws=None, lambda_draws=None, factor_draws=None, h_factor_draws=None, h0_factor_draws=None, sigma_eta2_factor_draws=None, gamma_draws=None, mu_draws=None, mu_gamma_draws=None)[source]

Bases: object

Output of srvar.api.fit().

Depending on the model configuration, this object may contain:

  • Closed-form NIW posterior parameters (posterior)

  • Stored posterior draws of coefficients/covariances (beta_draws, sigma_draws)

  • Latent shadow-rate series/draws when ELB is enabled

  • Stochastic volatility state draws when volatility is enabled

  • SSVS inclusion indicator draws when prior.family='ssvs'

beta_draws: ndarray | None
dataset: Dataset
factor_draws: ndarray | None
gamma_draws: ndarray | None
h0_draws: ndarray | None
h0_factor_draws: ndarray | None
h_draws: ndarray | None
h_factor_draws: ndarray | None
lambda_draws: ndarray | None
latent_dataset: Dataset | None
latent_draws: ndarray | None
property loading_draws: ndarray | None

Alias for factor SV loading draws (Lambda).

This mirrors lambda_draws (kept for backwards compatibility).

model: ModelSpec
mu_draws: ndarray | None
mu_gamma_draws: ndarray | None
posterior: PosteriorNIW | None
prior: PriorSpec
q_draws: ndarray | None
sampler: SamplerConfig
sigma_draws: ndarray | None
sigma_eta2_draws: ndarray | None
sigma_eta2_factor_draws: ndarray | None
sv_gamma0_draws: ndarray | None
sv_phi_draws: ndarray | None
class srvar.results.ForecastResult(variables, horizons, draws, mean, quantiles, latent_draws=None)[source]

Bases: object

Output of srvar.api.forecast().

draws: ndarray
horizons: list[int]
latent_draws: ndarray | None
mean: ndarray
quantiles: dict[float, ndarray]
variables: list[str]
class srvar.results.HistoricalDecompositionResult(variables, shocks, time_index, baseline_draws, shock_draws, contributions_draws, mean, quantiles, identification, metadata=<factory>)[source]

Bases: object

Historical decomposition (HD) draws and summaries.

This decomposes observed (or latent) series into a baseline path plus shock contributions implied by a structural identification scheme.

baseline_draws: ndarray
contributions_draws: ndarray
identification: str
mean: ndarray
metadata: dict[str, Any]
quantiles: dict[float, ndarray]
shock_draws: ndarray
shocks: list[str]
time_index: Any
variables: list[str]
class srvar.results.IRFResult(variables, shocks, horizons, draws, mean, quantiles, identification, metadata=<factory>)[source]

Bases: object

Impulse response function (IRF) draws and summaries.

draws: ndarray
horizons: list[int]
identification: str
mean: ndarray
metadata: dict[str, Any]
quantiles: dict[float, ndarray]
shocks: list[str]
variables: list[str]
class srvar.results.PosteriorNIW(mn, vn, sn, nun)[source]

Bases: object

NIW posterior parameter block.

This is returned in FitResult when the model is conjugate and closed-form NIW posterior parameters are available.

mn: ndarray
nun: float
sn: ndarray
vn: ndarray