Detect Recurring Transactions

Problem: Detect Recurring Transactions Given a transaction file, identify transactions that likely represent recurring charges or subscriptions. Each transaction contains: transaction_id, merchant, date, amount_cents where: transaction_id is a unique string; merchant is a merchant identifier; date is in YYYY-MM-DD format; amount_cents is a positive integer in cents. A set of transactions for the same merchant is considered recurring if it contains at least 3 transactions, with approximately stable gaps between consecutive transactions and approximately stable amounts. Tunable parameters The function receives: interval_tolerance_days: maximum allowed deviation of a consecutive date gap from the median interval; amount_tolerance_cents: maximum allowed deviation of a transaction amount from the group's median amount. Output Return every detected recurring transaction group. Each group contains: merchant, estimated_interval_days, median_amount_cents, transaction_ids where: estimated_interval_days is the median consecutive date gap, rounded to an integer; median_amount_cents is the median transaction amount, rounded to an integer; transaction_ids are ordered by date. Sort groups lexicographically by merchant. Example Input: 8 1,Netflix,2024-01-05,1599 2,Netflix,2024-02-05,1599 3,Netflix,2024-03-06,1699 4,Gym,2024-01-01,5000 5,Gym,2024-02-01,5000 6,Gym,2024-03-01,5000 7,Coffee,2024-01-01,500 8,Coffee,2024-01-03,500 With interval_tolerance_days = 3 and amount_tolerance_cents = 150, output: Gym,30,5000,4|5|6 Netflix,30,1599,1|2|3 Coffee is not recurring because it has only two transactions. Example Input 8 3 150 1,Netflix,2024-01-05,1599 2,Netflix,2024-02-05,1599 3,Netflix,2024-03-06,1699 4,Gym,2024-01-01,5000 5,Gym,2024-02-01,5000 6,Gym,2024-03-01,5000 7,Coffee,2024-01-01,500 8,Coffee,2024-01-03,500 Output Gym,30,5000,4|5|6 Netflix,30,1599,1|2|3