Frequent Itemset Mining Based on Heuristic Two Level Counting

Recently, many enchanced Apriori algorithms have been proposed to efficiently generate all frequent itemsets from datasets in data mining field. Although efficient techniques were presented, those algorithms are either time-consuming or memory-consuming. To address the issue further, a new algorithm, which introduced a novel support counting method, heuristic two level counting, is proposed. HTLC method adopts an improved itemset generating technology in the generation process of low level itemsets, which promotes the production of low level frequent itemsets or candidate itemsets. It also applies a heuristic traversal technology which speeds up one pass over datasets and support counting technology which largely reduces the number of passes over datasets to the generation of high level frequent itemsets. Finally, the experimental results show that it outperforms existing Apriori-like algorithms in mostly datasets