arviz_stats.base.array_stats.loo_approximate_posterior

array_stats.loo_approximate_posterior(ary, log_p, log_q, chain_axis=-2, draw_axis=-1, log_jacobian=None)

Compute PSIS-LOO-CV with approximate posterior correction.

Parameters:

aryarray_like

Log-likelihood values.

log_parray_like

Target log-density values.

log_qarray_like

Proposal log-density values.

chain_axisint, default -2

Axis for chains.

draw_axisint, default -1

Axis for draws.

log_jacobianfloat, optional

Log-Jacobian adjustment for variable transformations.

Returns:

elpd_iarray_like

Pointwise expected log predictive density.

pareto_karray_like

Pareto k-hat diagnostic values.

p_loo_iarray_like

Pointwise effective number of parameters.