Back to problems

Measure feature impact with switchback, PSM, and CACE

System Design · Uber · Hard

Imagine you are on a data science team at a ride-sharing platform. A new membership offering has been introduced, and leadership wants to understand how it changes rides per user (RPU). This exercise asks you to quantify that effect under three different evidentiary conditions: a switchback experiment, an observational launch without randomization, and a randomized experiment with imperfect membership take-up. For each approach, your answer should: name the identifying…

Checking your access…