1`8 - Apriori Rule Gen

Rule Generation

Generating association rules from list of frequent itemsets to find those that meet the minconf threshold

Confidence formula

Confidence Based Pruning

If the confidence of the rule is below the min threshold, then any rule with the same LHS and larger RHS will have a equal or lower confidence ( prune all subsequent rules )

Rule Generation Algorithm

Given a frequent itemset, generate candidate rules, compute confidence for each rule, keep rule if >= min conf, prune if < min conf

How to set up minsup?

Base it initially on the size of the dataset

How to set up minconf?

Base it on how much confidence is believed to demonstrate a relation between X and Y

How to tune minsup and minconf?

Based on the # of total rules generated after algorithm