Detect Duplicate Transactions in 5 Minutes

Write a query using window functions and conditional aggregations to identify potential duplicate transactions occurring within a 5-minute window.

Treat transactions with the same non-null account, merchant, amount, and currency as potential duplicates. Only completed transactions count toward duplicate detection, and the five-minute boundary is inclusive.

Output

1. Return one row per completed transaction that belongs to a potential duplicate group.
2. Include transaction_id, transaction attributes, previous_completed_transaction_id, and duplicate_count.
3. Order by occurred_at, then transaction_id.