association rules mining

The goal of this procedure it’s, given a list of $N$ item-set, finding association rules that have $conf$ and $sup$ grater than some thresholds

brute-force approach

generate all possible combination and compute $conf$ and $sup$, this approach is always possible but is too much computational expensive

two step approach

this approach is based on the fact that rules that are generated from the same item-set have the same $sup$

• frequent itemset generation -> in the first step all item-set that have $sup \gt threshold$ are generated (this step is still computational expensive)
• RULE GENERATION -> in the second step rules with high confidence are generated from the previous generated item-sets