Struct SyntheticControl 

pub struct SyntheticControl {
    pub max_iter: usize,
    pub tol: f64,
}

Synthetic Control Method (Abadie-Diamond-Hainmueller, 2010).

Finds a weighted combination of donor units that best matches the treated unit’s pre-treatment trajectory. The weights are constrained to be non-negative and sum to one; they are found by projected-gradient descent.

Fields

max_iter: usize

Maximum number of optimization iterations

tol: f64

Convergence tolerance

Implementations

impl SyntheticControl

pub fn new() -> Self

Create a new SyntheticControl estimator.

pub fn fit_weights( &self, y_treated: &ArrayView1<'_, f64>, y_donors: &ArrayView2<'_, f64>, ) -> StatsResult<Array1<f64>>

Fit the synthetic control weights.

Arguments

• y_treated – pre-treatment outcomes for the treated unit (T_pre,)
• y_donors – pre-treatment outcomes for donor units (T_pre × n_donors)

Returns

Optimal weights (n_donors,), sum to 1, all >= 0.

pub fn treatment_effects( &self, y_treated_post: &ArrayView1<'_, f64>, y_donors_post: &ArrayView2<'_, f64>, weights: &ArrayView1<'_, f64>, ) -> StatsResult<Array1<f64>>

Estimate treatment effect for each post-treatment period.

Arguments

• y_treated_post – post outcomes for treated unit (T_post,)
• y_donors_post – post outcomes for donor units (T_post × n_donors)
• weights – fitted weights from fit_weights

Trait Implementations

impl Default for SyntheticControl

fn default() -> Self

Returns the “default value” for a type.