Causal impact results

polars_ts.causal.causal_impact_results

CausalImpactResult dataclass — result container for causal impact analysis.

CausalImpactResult dataclass

Result container for a CausalImpact analysis.

All effect arrays have length equal to the post-intervention period.

Attributes

point_effect Pointwise causal effect (observed - counterfactual). point_effect_lower Lower credible bound of the pointwise effect. point_effect_upper Upper credible bound of the pointwise effect. cumulative_effect Cumulative sum of pointwise effects. cumulative_effect_lower Conservative lower bound of cumulative effect (sum of pointwise lower bounds). This is wider than a proper posterior-based interval because it does not account for cross-timestep correlation. cumulative_effect_upper Conservative upper bound of cumulative effect. total_effect Sum of pointwise effects over the post period. total_effect_lower Lower credible bound of total effect. total_effect_upper Upper credible bound of total effect. relative_effect Total effect divided by sum of counterfactual. relative_effect_lower Lower credible bound of relative effect. relative_effect_upper Upper credible bound of relative effect. counterfactual Predicted counterfactual series for the post period. counterfactual_lower Lower credible bound of counterfactual. counterfactual_upper Upper credible bound of counterfactual. observed_post Observed values in the post period. bsts_result Underlying BSTS model result. pre_mape Mean absolute percentage error on the pre-period (diagnostic). pre_coverage Fraction of pre-period observations inside the credible interval.