What Is Sequence Pattern Mining?

You might have heard of association rule mining (ARM) which allows you to generate association rules to display the relationships between items in a dataset.

However, one can apply Sequence pattern mining to look for time-ordered association patterns.

For example, A customer of an eCommerce websites makes the following purchase

From the table above, one can see that the purchase results of Cust1 results in a sequence of item purchased based on the purchase date.

The main difference between Sequence pattern mining and ARM is that the time sequence is taken into account. We are also studying associations that occur over a period of time.

Sequence pattern mining is applicable for datasets that have a time-related format and can be used to develop marketing or campaign strategies.

Development of Sequence Rules

Compared to traditional antecedents and consequents that are both inputs and outputs in an Apriori algorithm, the itemset leading to the final Itemset forms the antecedent sequence while the last item forms the consequent sequence.