Define and estimate prevalence of unhealthy users

Using the same Oculus tables (`user_activity`, `apps`), you’re asked: ## Question “How common are **unhealthy users**?” ## Task 1) Propose a concrete, measurable definition of an “unhealthy user” using session `duration` and dates (e.g., heavy usage thresholds across multiple days). 2) Describe how you would compute the **prevalence** (percentage of users who are unhealthy) over the last 30 days. 3) List 2–3 pitfalls/data issues you’d check (e.g., outliers, timezone/day-boundary, missing duration).