Implementation of RP-growth in Python from the paper "Discovering Recurring Patterns in Time Series"

1. Create a Transaction Database from your data. Transaction databases have the following format: tdb = { 1: ["a", "b", "g"], 2: ["a", "c", "d"], 3: ["a", "b", "e", "f"], 4: ["a", "b", "c", "d"], 5: ["c", "d", "e", "f", "g"], 6: ["e", "f", "g"], 7: ["a", "b", "c", "g"], 9: ["c", "d"], 10: ["c", "d", "e", "f"], 11: ["a", "b", "e", "f"], 12: ["a", "b", "c", "d", "e", "f", "g"], 14: ["a", "b", "g"], }

• Each row represents a transaction which has a timestamps and events/items that occurred in this transaction
• Transactions need to be equally spaced, e.g., every month/day/hour
• Empty transactions are skipped
• Items/Events need to be hashable

2. Create an instance of the PatternFinder class pattern_finder = PatternFinder(tdb, per, min_ps, min_rec)

• tdb: the transaction database
• per (period): maximum distance between two consecutive occurrences of an item to be counted periodic
• minPS (minimum support): minimum number of items in a given recurrence
• minRec (minimum recurrence): minimum number of recurrences

3. Get the patterns patterns = pattern_finder.find_recurring_patterns()

This repository is based on the following paper: Kiran, R., Shang, H., Toyoda, M., & Kitsuregawa, M. (2015). Discovering Recurring Patterns in Time Series. 18th International Conference on Extending Database Technology. ISBN:978-3-89318-067-7

Link: https://openproceedings.org/2015/conf/edbt/paper-23.pdf