Usage of sequential patterns

Usually, the market basket analysis runs on a large fact table or on a part of it.

Such a table contains at least the following columns:
  1. A sequence ID, such as a user ID or credit card number
  2. A transaction ID column or time ID column
  3. An item ID column
The fact table contains each transaction, that is, each purchase order, with the purchased items. Moreover, each transaction is associated with a sequence. To save disk usage, the columns are often numeric. Additionally, the relation between the item ID and the name or description of the purchased item can be stored in a separate table. The transactions of a sequence in such a table are sorted by transaction ID or time ID.

The sequential patterns analysis can find frequent sequential patterns in this table, that is, sequences of purchase patterns. Based on the findings, the sequential patterns analysis can imply sequential rules.

(A)(B)(C)
Indicates that many customers purchase item A in one transaction, item B in a subsequent transaction, and item C in a later transaction.
(A B)(C)
Indicates that many customers purchase items A and B in the same transaction, and item C in a later transaction.