Design causal study for airport cancellation reduction

You are a Senior Data Scientist supporting an airport pickups team at a rideshare company. Context: - Airport pickups are operationally different from city pickups because drivers often wait in an **airport queue**. - A driver cancellation or rider cancellation can be especially costly: drivers may lose their place in the queue and/or have to leave and re-enter, harming driver experience. - Riders at airports may be confused about the correct pickup location and may include many one-time users; trip intent (business vs personal) may also differ. - There may be **network effects/interference**: changing dispatch, pricing, or guidance for some users can affect others waiting in the same queue. Goal: Reduce the airport trip **cancellation rate** while not harming marketplace health. Questions: 1) **Problem framing & metrics**: Define supply, demand, and marketplace health for airport pickups. Propose a primary metric (or a small set) for “reduce cancellations” plus guardrails. Be explicit about definitions (e.g., what counts as a cancellation; time windows). 2) **Hypotheses**: List plausible, testable hypotheses for why cancellations occur at airports for (a) drivers and (b) riders. 3) **Causal identification / experimentation design**: Airports are hard to test with standard geo experiments; simple Diff-in-Diff may fail due to strong time-varying confounding (flight schedules, weather, events). Propose one or more practical designs to estimate causal impact of an intervention to reduce cancellations, addressing: - interference/network effects from a shared airport queue - time variation (seasonality, flight arrival waves) - operational constraints (limited number of airports; cannot “turn off” the feature for an entire region) 4) **Satisfaction / happiness measurement**: Propose data-driven proxy metrics for **driver satisfaction** and **rider satisfaction** specific to airports, and explain tradeoffs/limitations. Also propose at least one non-data-first (behavioral/psychology) hypothesis you would investigate. Assume you have access to standard marketplace logs (trip lifecycle events, dispatch events, queue position changes, app events), driver/rider attributes, and airport identifiers.