Top K Active Users

First clarify what activity means and whether repeated events, bot events, and time-zone boundaries count. Aggregate events by user id for the requested window. If all counts fit in memory, sorting users by count descending and the stated secondary key is simple. For a streaming or very large input, keep a min-heap of size k keyed by count and tie breaker; push each aggregated user and discard the smallest when size exceeds k. The heap approach costs O(E + U log k) after aggregation, versus O(E + U log U) for sorting, with O(U) map storage. In SQL use GROUP BY user_id, COUNT(*), then ORDER BY count DESC and LIMIT k. Test fewer than k users, tied counts, empty activity, and a user with many duplicate events.