Outline

1. generate candidates of size k = 1
   1. create 1 itemset for each regular item

   2. create event items and itemsets for them
      1. for each event type in the data set
         1. repeat the following the minumum of
            • the maximum number of same type event items specified by the user
            • the number of occurrances of that type in the data set

         2. create a new event item of that type with begin 0 and end time 1 and add it to the list of items

         3. create a candidate itemset containing the new event item

2. count support of candidate itemsets of size k = 1

3. add itemsets with support above minimum to frequent itemsets of size k list

4. generate candidates of size k = 2

5. count support of candidate itemsets of size k

6. add itemsets with support above minimum to frequent itemsets of size k list

7. generate candidates of size k + 1

8. continue repeating steps 5, 6, and 7 until no more candidates can be generated

9. generate association rules from frequent itemsets