This problem requires a Pro subscription. Upgrade to access all 15 industry domains and 1,500+ questions.
Upgrade to ProYou are a Staff Experimentation Scientist at Meta. Meta's Feed team runs CUPED (Controlled-experiment Using Pre-Experiment Data) on every A/B test to reduce variance. Idea: if you know a user's pre-experiment activity, you can adjust their post-experiment outcome by that covariate and dramatically tighten the confidence interval. Demonstrate this on a synthetic A/B assignment by parity of user_id.
You are a Staff Experimentation Scientist at Meta. Meta's Feed team runs CUPED (Controlled-experiment Using Pre-Experiment Data) on every A/B test to reduce variance. Idea: if you know a user's pre-experiment activity, you can adjust their post-experiment outcome by that covariate and dramatically tighten the confidence interval. Demonstrate this on a synthetic A/B assignment by parity of user_id.