針對時空性交易資料提供動態視窗關聯法則探勘之研究

傳送到

題名:
針對時空性交易資料提供動態視窗關聯法則探勘之研究
著者: 陳柏宏; Chen, Po-hung
李建億; Chien-I Lee; 數位學習科技學系
主題: 購物動線; 時空性關聯法則; 時空性資料探勘; 資料探勘; shopping flow; spatio-temporal data mining; STARs; data mining
描述: 在傳統的購物探勘上，只能檢視顧客最後結帳時所購買的商品，而無法了解顧客購物時的購物動線。關聯法則（Association Rules）由賣場的交易資料庫中挖掘顧客的購買行為，探勘頻繁商品集合，亦只著重購買商品之間的關係。現今隨著科技的進步，不僅能在結帳時由收銀台取得顧客購買商品的交易資訊，在購物車設置RFID或其他感應裝置，隨著購物動線，還可以額外取得商品放入購物車的時間和位置。購物動線包含了傳統交易所沒有的時間與空間資訊。有了這些額外的資訊，可能對這些有興趣：顧客的購物動線、顧客在一定範圍內會一起購買哪些商品、顧客在一定時間內會一起購買哪些商品。因此，為了解決傳統關聯法則不支援對時間與空間多重維度之分析，本研究提出支援動態視窗之時空性交易關聯法則（Spatio-Temporal Transaction Association rules with the Dynamic Window, STTA/DW）。以Apriori演算法為基礎，支援對時間與空間多重因素之分析；並提供動態視窗在改變探勘條件時，以參考同一父交易的其他子交易來取代重新掃描整個資料庫，加速重新探勘的速度。實驗結果顯示，本研究在不同條件之下，能有效地正確地探勘出具有時間與空間的時空性關聯法則，並且也比使用傳統Apriori探勘出的結果更加快速且準確。
We can only check what the customers buy on the traditional data mining for shopping, but we can not understand the shopping flow when the customers shop. Association Rules know the shopping behavior of customers from transactions database in the market, search for frequent item ets, but only pay attention to the relationship between the items.Nowadays along with technical progress, we can not only obtain the transaction information of the customers’ purchasing items when paying up by cash register, but also extra obtain when and where items put in the shopping vehicle along with the shopping flow by establishing RFID or other sensor devices in the shopping vehicle. The shopping flow contains the time and location information, but the traditional transaction does not. With the extra information, we are possibly interested in the follows: Customers’ shopping flow, which items together can the customer purchase in the certain scope, and which items together can the customer purchase in the certain time.Therefore, in order to solve the traditional association rules which has not support the analysis of the time and spatial multiple dimensions, this research proposes Spatio-Temporal Transaction Association rules with the Dynamic Window (STTA/DW). Based on the Apriori algorithm, it supports analysis of the time and spatial multiple factor, and also provides the Dynamic Windows for modification of parameters of mining. When parameters are modified, it refers to other sub-transactions which have the same parent, instead of re-scanning the whole database. The experimental result showed that this research under the different conditions not only can mining out spatio-temporal rules effectively and correctly, and is more quick and exact than traditional Apriori.
碩士
建立日期: 2007
格式: 121 bytes
text/html
語言: 中文
識別號: http://nutnr.lib.nutn.edu.tw/handle/987654321/3907
資源來源: NUTN IR

連結

連回原始介面

針對時空性交易資料提供動態視窗關聯法則探勘之研究

陳柏宏; Chen, Po-hung 李建億; Chien-I Lee; 數位學習科技學系 2007

正在檢索遠程資料庫，請稍等