ModelEngine

models.ModelEngine()

Anything that turns parameter draws into standardized outcomes.

Implementations must satisfy two invariants:

1. The returned Outcomes iteration index equals draws.index (same values, same order); this preserves the parameter/outcome linkage required by EVPPI and EVSI.
2. Every intervention is evaluated on every iteration (a balanced panel); the Outcomes constructor enforces this.

Example

import pandas as pd from heormodel.models import ModelEngine, Outcomes class TwoInterventionModel: … def evaluate(self, draws: pd.DataFrame) -> Outcomes: … costs = pd.DataFrame({“A”: draws[“c”], “B”: draws[“c”] * 2}) … effects = pd.DataFrame({“A”: draws[“e”], “B”: draws[“e”] * 1.5}) … return Outcomes.from_wide(costs, effects) isinstance(TwoInterventionModel(), ModelEngine) True

Methods

Name	Description
evaluate	Evaluate all interventions for every row of the draw matrix.

evaluate

models.ModelEngine.evaluate(draws)

Evaluate all interventions for every row of the draw matrix.