Skip to content
ML Academy

Statistics

Interview question

  • midlevel
  • 10 minsuggested time
  • 2Data scientist, Data analyst

A league table ranks two regional carriers on on-time performance (OTP), defined as the share of operated flights that arrive within 15 minutes of schedule (hypothetical figures). Carrier X: 1,000 scheduled flights, 100 cancelled, 810 of the 900 operated flights on time, so OTP 90.0%. Carrier Y: 1,000 scheduled flights, 10 cancelled, 861 of the 990 operated flights on time, so OTP 87.0%. A journalist writes "X is clearly the more punctual carrier". Is that supported, and what would you report?

Tests: Selection on survival: how dropping cancelled (never observed) units from a denominator biases a rate, when a ranking is not identified from the data, bounds and break-even assumptions, and choosing a metric definition that matches the decision.

day 5day 6

Sign in to practise: answers are saved before model answers are shown.