RevenueRegressionScenario

Economics scenarios · Implementation: mlsysim/engine/scenario_definitions/economics.py.

Construct this scenario with keyword inputs. The fields below show the types and defaults declared by this class.

A metric regression that cuts revenue in proportion for undetected (or for alerted with an alert and containment); revenue is weekly_revenue per week.

Notes

Equation: Loss is relative_drop*weekly_revenue*(duration/one_week); shorter windows reduce it proportionally.

Assumptions: Constant revenue rate and a linear assumed revenue response to the metric regression.

Excludes: Demand elasticity, recovery lag, and causal identification of the metric-revenue relation.

References

Derivation: The scenario assumes revenue falls in direct proportion to relative_drop. A constant weekly revenue rate integrated over duration therefore loses relative_drop times revenue_rate times duration; an earlier alert changes only the exposure window.

Examples

>>> from mlsysim import Scenarios
>>> s = Scenarios.AggregationWindowRevenueIncident
>>> s.undetected_loss.to("USD").magnitude >= s.alerted_loss.to("USD").magnitude
True

Attributes

Name	Description
alerted
alerted_loss	Revenue loss in USD over the alert-and-containment window.
feature_mean_after
feature_mean_before
feature_shift	Signed relative shift of the feature mean from its initial value.
freshness_slo
impact_reduction	Dimensionless fraction of proportional loss avoided by the shorter detection window.
metric_after
metric_before
metric_change	Signed relative change of the monitored metric from its initial value.
relative_drop
shift_threshold
undetected
undetected_loss	Revenue loss in USD over the uncontained incident window.
weekly_revenue
window_ratio	Dimensionless uncontained duration divided by the contained duration.

Methods

Name	Description
loss	Revenue lost in USD over the supplied duration at the assumed proportional drop.

loss

engine.scenario_types.RevenueRegressionScenario.loss(duration)

Revenue lost in USD over the supplied duration at the assumed proportional drop.

Equation: (relative_drop * weekly_revenue * (duration / Q_(1, 'week')).to('').magnitude).to('USD').