Skip to content
voiage

voiage.ml_policy_voi.compute_policy_uplift_voi

compute_policy_uplift_voi

Section titled “compute_policy_uplift_voi”
compute_policy_uplift_voi([positional or keyword] cate_samples: np.ndarray = None, [positional or keyword] intervention_cost: float = None, [positional or keyword] payoff_multiplier: float = None, [positional or keyword] budget_constraint: float | None = None, [positional or keyword] subgroups: dict[str, np.ndarray] | None = None, [positional or keyword] evaluation_id: str = 'eval_uplift_01') -> UpliftVOIResult

Compute heterogeneous policy and uplift Value of Information.

Parameters

Section titled “Parameters”

cate_samples : np.ndarray 2D array of shape (n_simulations, n_units) containing posterior CATE draws (treatment effect on event probability or outcome). intervention_cost : float Unit cost to apply intervention. payoff_multiplier : float Value generated per unit of treatment effect (e.g. CLV or saved cost). budget_constraint : float, optional Maximum total spend allowed across units. subgroups : dict[str, np.ndarray], optional Boolean masks of shape (n_units,) defining named customer segments. evaluation_id : str Unique evaluation identifier.

Returns

Section titled “Returns”

UpliftVOIResult Expected net benefit under current policy, optimal policy, and uplift EVPI.

Parameters:

  • cate_samples np.ndarray
  • intervention_cost float
  • payoff_multiplier float
  • budget_constraint float | None (default: None)
  • subgroups dict[str, np.ndarray] | None (default: None)
  • evaluation_id str (default: 'eval_uplift_01')

Returns: UpliftVOIResult