Skip to content
ML Academy

S17 · company simulation

A vendor no-show model trial that wants the whole booking record

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

  • 4stages
  • 40 minabout
  • 0stages committed

The brief

A vendor (simulated) offers a 90-day trial of a no-show prediction service for the synthetic airline and asks for 24 months of passenger-level booking records so it can "train a custom model". The airline operates 19 flights a day (6,935 departures a year, as in the course database) and about 1.1 million bookings a year (1.07 to 1.13 million in the course database). You currently use an in-house logistic model. All people, opinions, fees and values are simulated or hypothetical; the passenger data is fictional.

Colleagues involved: a product manager who wants the trial in four weeks; a privacy and legal reviewer; a security lead; a data engineer; a finance partner; and the commercial director.

Sources used here: the principles of the EU GDPR text (source: docs-eu-gdpr-2016-679; this is not legal advice and other jurisdictions differ) and NIST guidance on protecting personally identifiable information (source: docs-nist-sp-800-122).

Stage 1 of 4

The vendor's request lists these fields per booking: name, e-mail, phone, passport number, payment token, IP address, booking key, fare class, days to departure, party size, sales channel, route, exact departure date, and the no-show outcome for every booking.

Which items may go in the first release to the vendor? Select all that apply.

Sign in to play scenarios (answers are recorded before the reveal).