Struct CounterfactualQuery 

pub struct CounterfactualQuery {
    pub scm: StructuralCausalModel,
}

Compute counterfactual queries of the form E[Y | do(X=x), Z=z].

Uses the abduction-action-prediction three-step procedure on a linear SCM.

Fields

scm: StructuralCausalModel

The structural causal model for the query.

Implementations

impl CounterfactualQuery

pub fn new(scm: StructuralCausalModel) -> Self

Create a new counterfactual query engine.

pub fn query_do( &self, target: usize, x_val: f64, outcome: usize, noise_samples: &[Vec<f64>], ) -> f64

Compute E[Y | do(X_target=x_val)] using the interventional distribution.

Arguments

• target — the variable to intervene on
• x_val — the intervention value
• outcome — the outcome variable
• noise_samples — noise samples for Monte Carlo evaluation

pub fn counterfactual( &self, obs: &[Option<f64>], target: usize, x_val: f64, outcome: usize, ) -> f64

Compute the counterfactual: given that we observed obs (variable→value pairs), what would outcome have been if do(target=x_val)?

Three steps:

1. Abduction: infer noise U from observations
2. Action: intervene on target
3. Prediction: compute outcome under intervention

Arguments

• obs — observed values, obs[v] = Some(value) if observed
• target — intervention variable
• x_val — intervention value
• outcome — outcome variable index

pub fn probability_of_necessity( &self, treatment: usize, outcome: usize, t_val: f64, t_counter: f64, threshold_y: f64, noise_samples: &[Vec<f64>], ) -> f64

Compute the probability of necessity (PN): P(Y=0 | do(X=0), Y=1, X=1).

Simplified calculation using noise samples.

Trait Implementations

impl Clone for CounterfactualQuery

fn clone(&self) -> CounterfactualQuery

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for CounterfactualQuery

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.