Struct XLearner 

pub struct XLearner { /* private fields */ }

X-learner (Künzel et al., 2019).

1. Fit μ₀, μ₁ per arm (outcome learner).
2. Impute effects: Y − μ₀(X) for treated, μ₁(X) − Y for controls.
3. Model the imputed effects per arm (effect learner): τ₁, τ₀.
4. Blend: τ̂(x) = e(x)·τ₀(x) + (1 − e(x))·τ₁(x) with propensity e.

Implementations

impl XLearner

pub fn new( outcome: impl Regressor + 'static, effect: impl Regressor + 'static, propensity: impl Classifier + 'static, ) -> Self

pub fn from_boxed( outcome: Box<dyn Regressor>, effect: Box<dyn Regressor>, propensity: Box<dyn Classifier>, ) -> Self

pub fn fit(&self, data: &DmlData) -> Result<CateFit>