高效率Apriori演算法探勘關聯規則

A High Efficient Apriori Algorithm of Mining Association Rules

探勘關聯規則（association rule）是資料探勘領域中一個相當重要的研究問題之一，本研究以交易資料為探勘資料來源，每一筆交易資料包含消費者曾經購買的產品項目，設計一個HE_Apriori演算法探勘關聯規則，其縮減組合形成新項目集時必須掃瞄的高頻項目集數量，且減少交易資料包含的項目個數，只包含高頻項目集的項目，並避免掃瞄未包含項目集的交易資料。HE_Apriori演算法在減少掃瞄高頻項目集數量、減少掃瞄交易資料數量及其包含之項目個數的情況下，達到提升探勘關聯規則之效能的目的。從實驗評估中顯示，HE_Apriori演算法的執行效率優於Apriori演算法探勘關聯規則。

In the field of data mining, mining association rules from the transaction database is one of the most popular problems. This paper uses transaction data as the source of mining, and each transaction data contains a consumer ever bought product items. An algorithm, called HE_Apriori is proposed to mine association rules. The algorithm reduces the amount of scanning frequent itemsets to generate new itemsets, the amount of scanning the transaction data which only contain frequent 1-itemset to generate frequent itemsets, and avoids scanning the transaction data which does not contain the itemsets. Following the above process can reduce the amount of scanning data. The experiments show that the HE_Apriori algorithm can effectively improve the performance of the Apriori algorithm for mining association rules.