SamplesSummary.aggregate_diagnostics()

SamplesSummary.aggregate_diagnostics()#

SamplesSummary.aggregate_diagnostics(by='min/max')[source]#

Aggregates effective sample sizes (ESS) and rhat.

Parameters:

by (Literal['min/max', 'mean', 'median', 'std', 'var', 'min', 'max'], default: 'min/max') –

How to aggregate. The three current options are:

  • "min/max": Aggregate ESS by taking the minimum ESS per parameter block, and the rhat by taking the maximum rhat per parameter block. This corresponds to a worst-case summary.

  • "mean": Aggregate ESS and rhat by averaging inside parameter blocks.

  • "median": Aggregate ESS and rhat by taking the median inside parameter blocks.

  • "std": Aggregate ESS and rhat by taking the standard deviation inside parameter blocks.

  • "var": Aggregate ESS and rhat by taking the variance inside parameter blocks.

  • "min": Aggregate ESS and rhat by taking the minimum inside parameter blocks.

  • "max": Aggregate ESS and rhat by taking the maximum inside parameter blocks.

Return type:

DataFrame

Notes

If per_chain is True, rhat cannot be computed and there is not present in the output dataframe.