Struct RdFlex 

pub struct RdFlex { /* private fields */ }

RD with flexible covariate adjustment. A fuzzy design (RdFlex::fuzzy) also adjusts the treatment with a classifier.

Implementations

impl RdFlex

pub fn sharp(ml_g: impl Regressor + 'static) -> Self

Sharp design: treatment is 1{score ≥ cutoff}. ml_g learns the outcome and must support sample weights.

pub fn fuzzy( ml_g: impl Regressor + 'static, ml_m: impl Classifier + 'static, ) -> Self

Fuzzy design: ml_m (supporting sample weights) learns the treatment.

pub fn from_boxed( ml_g: Box<dyn Regressor>, ml_m: Option<Box<dyn Classifier>>, ) -> Self

ml_m = Some(_) fits a fuzzy design.

pub fn sharp_binary(ml_g: impl Classifier + 'static) -> Self

Sharp design with a binary (0/1) outcome learned by the classifier ml_g (its class-1 probability), as DoubleML with a classifier.

pub fn fuzzy_binary( ml_g: impl Classifier + 'static, ml_m: impl Classifier + 'static, ) -> Self

Fuzzy design with a binary outcome learned by the classifier ml_g.

pub fn from_boxed_binary( ml_g: Box<dyn Classifier>, ml_m: Option<Box<dyn Classifier>>, ) -> Self

Binary outcome with a boxed classifier; ml_m = Some(_) fits a fuzzy design.

pub fn rd_options(self, options: RdOptions) -> Self

Options of the RD estimator (rdrobust’s keyword arguments in DoubleML: polynomial order, kernel, bandwidth selector, variance, …). The initial first-stage bandwidth is always selected with the defaults.

pub fn rd_covariates(self, covs: Array2<f64>) -> Self

Covariates for the RD estimator itself (rdrobust’s covs), one row per observation; the first stage uses the data’s covariates.

pub fn rd_clusters(self, labels: Vec<i64>) -> Self

Cluster labels for cluster-robust RD standard errors (rdrobust’s cluster); the sample splits are not clustered, as in DoubleML.

pub fn cutoff(self, cutoff: f64) -> Self

pub fn h_fs(self, h: f64) -> Self

Initial first-stage bandwidth (default: rdbwselect without covariates).

pub fn first_stage(self, spec: FirstStage) -> Self

pub fn kernel(self, kernel: RdKernel) -> Self

pub fn n_iterations(self, n: usize) -> Self

Rounds of first-stage fitting and bandwidth updating (default 2).

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

Number of cross-fitting folds (default 5).

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

Number of independent sample splits (default 1), aggregated by the median.

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

Seed for the sample splits (default 0).

pub fn fit( &self, data: &DmlData, score: ArrayView1<'_, f64>, ) -> Result<RdFlexFit>

Fit with sample splits stratified by the treatment; score is the running variable.

pub fn fit_with_folds( &self, data: &DmlData, score: ArrayView1<'_, f64>, splits: &[Vec<Fold>], ) -> Result<RdFlexFit>

Fit on given sample splits (one Vec<Fold> per repetition).