複合式關聯式法則運用於保險業資料採礦之研究

在資料採礦中，關聯式法則是經常被使用的技術之一，然而對於新上市的產品而言，關聯式法則的運用卻受到支持度及信賴區間最小門檻值的限制。因此，本文提出一個可以提昇支持度及信賴區間的演算法，藉此將多個項目合成一個複合式物品項目，產生以此複合式物品項目為後項的複合式關聯法則；並將此法則運用於保險業的產品組合及行銷。由實證結果顯示，主要險種搭配特定的附險銷售時，消費者除了主險外也會一併購買附險。
關鍵字：資料採礦、關聯性法則、複合式物品項目

Data Mining － Disjunctive Consequent Association Rule Research in Insurance Industry

　　　　　　　　　　　　　　　(Ding-An Chiang)*(Yi-Jen Hu)*(Cheng-Jung Lin)

The association rule is one of the frequently adopted techniques in data mining. However, it practically limited to the minimum support and confidence for newly marketed products. Therefore, we propose a new algorithm to discover the disjunctive consequent association rules whose consequents are formed by a set of disjunctive items. Moreover, we apply this new algorithm to cross selling in the insurance. The results appeared to customers would accept the primary insurance and other policies when they were provided with more than one specific insurance policies.
Keywords: data mining, association rule, composite item