S17 · company simulation

Run an overbooking model through the whole company

People, data and outputs in this scenario are simulated. Each stage is recorded when you commit; the reveal comes after.

5stages
45 minabout
0stages committed

The brief

You own a project at the synthetic airline: a model that recommends an overbooking limit per flight from forecast no-shows, replacing a fixed 6% rule. Seven colleagues are involved. Every person, opinion and number below is simulated; prices and costs are hypothetical.

Product manager: cares about agent workflow and launch date.
Data engineer: owns the booking and check-in pipelines.
Data scientist: built the no-show model.
ML engineer: owns serving, latency and rollback.
Security and privacy lead: owns access to passenger data.
Finance partner: approves the business case.
Executive sponsor (commercial director): decides go or no-go.

Stage 1 of 5

Each colleague tells you what would convince them. You need to plan which evidence goes to whom.

Which pairings of role and evidence standard are appropriate? Select all that apply.

Finance (simulated) needs expected value with a range and its assumptions, including denied-boarding costs.The ML engineer (simulated) needs service latency, failure behaviour, a rollback path and versioned artefacts.The executive (simulated) needs the full feature-importance table to approve, so the decision is defensible.Security (simulated) needs a data-flow diagram of passenger fields used, where stored and who has access.The data engineer (simulated) needs only the model's AUC, since the rest of the pipeline is the modelling team's concern.