Design and Evaluate an Experiment on Surge

Uber proposes a new surge-cap algorithm intended to improve trip completion in NYC 20:00–22:00 without materially harming ETA or fairness. Given baselines (historical for NYC, 20:00–22:00): - Requests per evening ≈ 500,000; baseline completion rate = 82%. - Average ETA = 6.5 minutes (SD across geo-cells = 2.0 minutes). - 250 geo-cells used for pricing; average requests per cell per evening ≈ 2,000; coefficient of variation across cells ≈ 0.6; intra-cell ICC for completion ≈ 0.05. Tasks: 1) Choose an experimental design (e.g., rider-level randomization, driver-level randomization, geo switchback by time, or city-level holdout). Justify your choice considering interference/spillovers in pricing and supply. 2) Define: primary metric, directional hypothesis, and guardrails (e.g., ETA, driver cancellations, demand spillover to adjacent hours, neighborhood fairness across income quartiles). Specify exact uplift and guardrail thresholds that would trigger a stop. 3) Power and duration: Target an MDE of +0.8 percentage points in completion. Using the provided aggregation structure (geo-cell as the unit) and ICC, approximate the number of switchback days needed to reach 80% power at α=0.05. State all assumptions (variance model, independence between alternating periods, balancing by weekday) and show the formula you use. 4) Diagnostics: Describe how you would run parallel-trend checks with pre-period data and a difference-in-differences estimator if you also include an external control market (e.g., Chicago). Specify the DID regression you’d run and the key robustness checks (placebo tests, event-study, cluster-robust SEs). 5) Implementation: Outline assignment, logging, and exposure rules to prevent cross-over contamination for riders and drivers during the test window.