metacausal.aggregation.EnsembleWeights¶

class metacausal.aggregation.EnsembleWeights(weights, model_names, intercept=0.0, method='', details=None)[source]¶

Bases: object

Result of weight computation for any non-pointwise aggregation.

Variables:

weights (numpy.ndarray) – Per-component weights, shape (K,). Sum to 1 for agreement and DR simplex strategies; non-negative for R-Stacking.
model_names (list[str]) – Adapter names in the same order as weights.
intercept (float) – Constant CATE shift. Nonzero only for R-Stacking.
method (str) – Which strategy produced these weights.
details (dict | None) – Method-specific metadata (e.g., mean_taus for CBA).

Parameters:

weights (ndarray)
model_names (list[str])
intercept (float)
method (str)
details (dict | None)

Methods

`__init__`
`plot`	Bar chart of these aggregation weights.

Attributes

`details`
`intercept`
`method`
`weights`
`model_names`

Details

plot(*, ax=None, on_uniform='warn', sort=True)[source]¶

Bar chart of these aggregation weights.

Thin wrapper around metacausal.plots.weights(); see there for the full parameter reference. Requires the plots extra (pip install 'metacausal[plots]').

Returns:

matplotlib Axes the plot was drawn on.

Parameters:

ax (Axes | None)
on_uniform (Literal['warn', 'error', 'ignore'])
sort (bool)

Return type:

Axes

details: dict | None = None¶

intercept: float = 0.0¶

method: str = ''¶

model_names: list[str]¶

weights: ndarray¶