rbartpackages.bartMachine.get_sigsqs

rbartpackages.bartMachine.get_sigsqs(bart_machine, after_burn_in=True, plot_hist=False, plot_CI=0.95, plot_sigma=False, verbose=True)[source]

Posterior draws of the error variance (regression fits only).

Burn-in draws are dropped unless after_burn_in=False, which also keeps the pre-MCMC initial value as first entry.

Parameters:

  • bart_machine (bartMachine) – The fitted model to read the draws from.

  • after_burn_in (bool, default: True) – Whether to drop the burn-in draws (and the pre-MCMC initial value).

  • plot_hist (bool, default: False) – Whether to plot a histogram of the post-burn-in draws.

  • plot_CI (float, default: 0.95) – Credible-interval level marked on the histogram (with plot_hist).

  • plot_sigma (bool, default: False) – Whether the histogram is of the SD rather than the variance (with plot_hist).

  • verbose (bool, default: True) – Whether to print messages to the R console.

Returns:

Float64[ndarray, 'num_iterations_after_burn_in'] | Float64[ndarray, 'num_gibbs+1'] – The error-variance draws (the pre-MCMC initial value first when after_burn_in=False).