SAPNSP: Select actionable positive and negative sequential patterns based on a contribution metric

摘要

Negative sequential patterns refer to sequences with non-occurring and occurring items, contrast to positive sequential patterns, its search space becomes very large. The number of these two kinds frequent sequence is also quite generous. However, not all of the discovered frequent patterns are useful to us, part of these sequences just occur frequently in database, but part of these sequences is actionable, because they conform to special rules rather than occur frequently. Here, we put forward a new method SAPNSP to select these actionable frequent positive and negative patterns. These selected sequential patterns are more conducive to our decision. Through real-world data and synthetic data to make experiments, we can see that the method is very effective to select these actionable patterns.