Can you list three reasons to enforce constraints (e.g., a minimal frequency constraint) when mining local patterns such as itemsets?
What is a closed itemset?, a maximal itemset? Give examples in a very small dataset.
What information is lost by condensing the frequent itemsets into the maximal itemsets? What about the closed itemsets?
Can you define the frequency of an itemset?, its lift? Illustrate those definitions with a very small dataset.
What is the practical advantage of a minimal frequency constraint that makes it more popular than a minimal lift constraint?
What is a maximal clique? How is it related to a closed itemset?
Can you define an association rule? What is its frequency?, its confidence? Some association rules can be the base of a classifier. Can you explain that?
