ConstraintAccuracyCostScenario
Statistics scenarios · Implementation: mlsysim/engine/scenario_definitions/statistics.py.
Construct this scenario with keyword inputs. The fields below show the types and defaults declared by this class.
Accuracy before and after a fairness constraint (percent), with the policy’s stated rates.
context holds the scenario’s descriptive rates in percent (default rates, approval rates, thresholds) so prose and slides quote one source.
Notes
Equation: Subtract constrained accuracy from baseline and normalize by baseline for relative cost.
Assumptions: Input accuracies use the same scale; percent inputs produce percentage-point differences.
Excludes: A causal price of fairness or a prediction outside the supplied comparison.
References
Derivation: Subtract constrained accuracy from baseline and normalize by baseline for relative cost. Individual property contracts identify the exact accounting expression or reusable calculation. Registry entries supply the scenario inputs; this class does not turn them into measured outcomes.
Examples
Evaluate a named scenario using its declared inputs.
>>> from mlsysim import Scenarios
>>> scenario = Scenarios.CreditParityAccuracyCost
>>> round(float(scenario.accuracy_cost), 4)
4.0Attributes
| Name | Description |
|---|---|
| accuracy_cost | Baseline minus constrained accuracy, in percentage points when input accuracies are |
| baseline_accuracy | |
| constrained_accuracy | |
| context | |
| relative_cost | Fraction of baseline accuracy lost under the supplied fairness constraint. |
baseline_accuracy
Type: float.
Declared default: required.
constrained_accuracy
Type: float.
Declared default: required.
context
Type: Dict[str, float].
Declared default: {}.
accuracy_cost
engine.scenario_types.ConstraintAccuracyCostScenario.accuracy_cost
Baseline minus constrained accuracy, in percentage points when input accuracies are percentages.
Points of accuracy given up.
Equation: baseline_accuracy - constrained_accuracy.
relative_cost
engine.scenario_types.ConstraintAccuracyCostScenario.relative_cost
Fraction of baseline accuracy lost under the supplied fairness constraint.
Accuracy given up as a fraction of the baseline.
Equation: accuracy_cost / baseline_accuracy.