metacausal.AteEstimate¶

class metacausal.AteEstimate(ate, component_estimates, aggregation, component_fit_times=None)[source]¶

Bases: object

Result from the ensemble of causal estimators.

Variables:

ate (float) – Aggregated treatment effect point estimate.
component_estimates (dict[str, metacausal.estimators.ComponentAteEstimate]) – Per-method estimates, keyed by method name.
aggregation (str) – Aggregation strategy class name.
component_fit_times (dict[str, float] | None) – Wall-clock fit time in seconds for each method on the full dataset. Does not include bootstrap iterations.

Parameters:

ate (float)
component_estimates (dict[str, ComponentAteEstimate])
aggregation (str)
component_fit_times (dict[str, float] | None)

Methods

`__init__`
`summary`	Return a formatted, multi-line ATE summary.

Attributes

`component_ates`	ATEs from each component method.
`component_fit_times`
`n_methods`	Number of methods that produced estimates.
`spread`	Range (max - min) of component ATEs.
`ate`
`component_estimates`
`aggregation`

Details

summary(*, digits=None, signed=False, show_ci=True)[source]¶

Return a formatted, multi-line ATE summary.

When show_ci is False, the per-component native confidence intervals are omitted, leaving a clean point-estimate table (useful when component CIs are unavailable for some methods or when the CI story is told separately, e.g. via bootstrap()).

Parameters:

digits (int | None)
signed (bool)
show_ci (bool)

Return type:

str

aggregation: str¶

ate: float¶

property component_ates: dict[str, float]¶: ATEs from each component method.

component_estimates: dict[str, ComponentAteEstimate]¶

component_fit_times: dict[str, float] | None = None¶

property n_methods: int¶: Number of methods that produced estimates.

property spread: float¶: Range (max - min) of component ATEs.