dist_reg_mcmc()

Contents

dist_reg_mcmc()#

liesel.model.dist_reg_mcmc(model, seed, num_chains, apply_jitter=True)[source]#

Configures an EngineBuilder for a distributional regression model.

The EngineBuilder uses a Metropolis-in-Gibbs MCMC algorithm with an IWLSKernel for the regression coefficients and a GibbsKernel for the smoothing parameters for a distributional regression model.

Parameters:

  • model (Model) – A model built with a DistRegBuilder.

  • seed (int) – The PRNG seed for the engine builder.

  • num_chains (int) – The number of chains to be sampled.

  • apply_jitter (bool) – If False, disables the jittering of the initial distributions regardless of the configuration in the variables inference configuration (see MCMCSpec). (default: True)

Return type:

EngineBuilder

Contents