srvar.artifacts¶
- class srvar.artifacts.FitNPZ(dataset, posterior, beta_draws, sigma_draws, q_draws, latent_dataset, latent_draws, h_draws, h0_draws, sigma_eta2_draws, sv_gamma0_draws, sv_phi_draws, lambda_draws, factor_draws, h_factor_draws, h0_factor_draws, sigma_eta2_factor_draws, gamma_draws, mu_draws, mu_gamma_draws)[source]¶
Bases:
object-
beta_draws:
ndarray|None¶
-
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_draws:
ndarray|None¶
-
mu_draws:
ndarray|None¶
-
mu_gamma_draws:
ndarray|None¶
-
posterior:
PosteriorNIW|None¶
-
q_draws:
ndarray|None¶
-
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¶
-
beta_draws:
- srvar.artifacts.load_run_dir(out_dir, *, config_filename='config.yml', fit_filename='fit_result.npz')[source]¶
Load a
FitResultfrom a srvar run output directory.This function:
Loads the stored draws/state from
fit_result.npz.Reconstructs
ModelSpec,PriorSpec, andSamplerConfigfrom the savedconfig.yml(without re-loading the original CSV).
Notes
The returned object is suitable for downstream analysis (IRFs/FEVD/HD) and forecasting.
If the saved config and saved dataset are inconsistent (e.g., variable list changed), config parsing may fail.
- Return type: