Struct RLearner 

pub struct RLearner { /* private fields */ }

R-learner (Nie & Wager, 2021): with cross-fitted m̂(x) = E[Y | X] and propensity ê(x), minimise Σ (Ỹ − τ(X)·D̃)², i.e. a weighted regression of Ỹ/D̃ on X with weights D̃². The effect learner must support Regressor::fit_weighted.

Implementations

impl RLearner

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

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

pub fn n_folds(self, n_folds: usize) -> Self

pub fn seed(self, seed: u64) -> Self

pub fn clipping_threshold(self, t: f64) -> Self

Propensities are clipped to [t, 1 − t] (default 0.01).

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

pub fn fit_with_folds(&self, data: &DmlData, folds: &[Fold]) -> Result<CateFit>