Background of association rules
Association rules mining detects frequent patterns and rules in transactions. Well-known algorithms for association rules mining are Apriori or FP-Growth. For analytic stored procedures, the PrefixSpan algorithm is preferred due to its scalability.
The Apriori algorithm searches for frequent item sets by doing consecutive iterations. These iterations systematically consider item sets of increasing size. For large data sets or for a low threshold, however, the number of identified candidates might be overwhelming.
Other than the Apriori algorithm, the PrefixSpan algorithm searches the complete set of patterns but avoids the creation of unnecessary candidates. Moreover, sorting the items and prefix-projection reduce the size of projected databases significantly. These methods also lead to efficient processing.