def summary(self, prob=0.9):
"""
Prints a summary table displaying diagnostics of samples obtained from
posterior. The diagnostics displayed are mean, standard deviation, median,
the 90% Credibility Interval, :func:`~pyro.ops.stats.effective_sample_size`,
:func:`~pyro.ops.stats.split_gelman_rubin`.
:param float prob: the probability mass of samples within the credibility interval.
"""
summary(self._samples, prob=prob)
def get_summary_table(posterior,
sites,
player_names,
transforms={},
diagnostics=False,
group_by_chain=False):
"""
Return summarized statistics for each of the ``sites`` in the
traces corresponding to the approximate posterior.
"""
site_stats = {}
for site_name in sites:
marginal_site = posterior[site_name].cpu()
if site_name in transforms:
marginal_site = transforms[site_name](marginal_site)
site_summary = summary({site_name: marginal_site},
prob=0.5,
group_by_chain=group_by_chain)[site_name]
if site_summary["mean"].shape:
site_df = pd.DataFrame(site_summary, index=player_names)
else:
site_df = pd.DataFrame(site_summary, index=[0])
if not diagnostics:
site_df = site_df.drop(["n_eff", "r_hat"], axis=1)
site_stats[site_name] = site_df.astype(float).round(2)
return site_stats
def get_summary_table(posterior, sites):
"""
Return summarized statistics for each of the ``sites`` in the
traces corresponding to the approximate posterior.
"""
site_stats = {}
for site_name in sites:
marginal_site = posterior[site_name].cpu()
site_summary = summary({site_name: marginal_site},
prob=0.5,
group_by_chain=group_by_chain)[site_name]
if site_summary["mean"].shape:
site_df = pd.DataFrame(site_summary)
else:
site_df = pd.DataFrame(site_summary, index=[0])
site_stats[site_name] = site_df.astype(float).round(2)
return site_stats
def summary(self, p=0.9, verbose=False):
self.results = pd.DataFrame(summary(self.mcmc_res._samples, prob=0.9))
if verbose: print(self.results)
return self.results
