metacausal.aggregation.PointwiseStrategy¶

class metacausal.aggregation.PointwiseStrategy[source]¶

Bases: AggregationStrategy

Base class for strategies that apply a fixed statistical rule pointwise.

Subclasses implement a single aggregate() method that reduces along axis 0 using NumPy axis=0-aware operations (np.median, np.mean, etc.). Because those operations are rank-agnostic, the same implementation handles both the CATE path ((K, n) → (n,), the base contract) and the ATE-from-component-scalars path ((K,) → 0-d scalar) — the latter is a Pointwise-specific extension of the base aggregate contract. Pointwise strategies are stateless: no weights are computed at fit time.

Methods

`__init__`
`aggregate`	Reduce a `(K, n)` component-CATE matrix to a `(n,)` ensemble CATE.

Attributes

`ensemble_weights`	Ensemble weights, if applicable.
`strategy_family`

Details

property ensemble_weights: None¶: Ensemble weights, if applicable. None for pointwise strategies.

strategy_family: ClassVar[str] = 'pointwise'¶