Skip to content
voiage

voiage.methods.ambiguity_distribution_shift.value_of_ambiguity_distribution_shift

value_of_ambiguity_distribution_shift

Section titled “value_of_ambiguity_distribution_shift”
value_of_ambiguity_distribution_shift([positional or keyword] value_array: ValueArray = None, [positional or keyword] shift_weights: np.ndarray | list[list[float]] = None, [positional or keyword] strategy_names: list[str] | None = None, [positional or keyword] scenario_names: list[str] | None = None, [positional or keyword] scenario_probabilities: np.ndarray | list[float] | None = None, [positional or keyword] ambiguity_radius: float = 0.0, [positional or keyword] information_cost: float = 0.0) -> AmbiguityDistributionShiftResult

Calculate robust VOI under source-target distribution shift.

Each row of shift_weights reweights source samples into a plausible target or drift scenario. The baseline uses a radius-penalized maximin score; the information value is the expected value of choosing after the shift scenario is resolved, less acquisition cost. The method remains fixture-backed pending stable promotion evidence.

Parameters:

  • value_array ValueArray
  • shift_weights np.ndarray | list[list[float]]
  • strategy_names list[str] | None (default: None)
  • scenario_names list[str] | None (default: None)
  • scenario_probabilities np.ndarray | list[float] | None (default: None)
  • ambiguity_radius float (default: 0.0)
  • information_cost float (default: 0.0)

Returns: AmbiguityDistributionShiftResult