應用資料探勘技術於農產品消費性電子商務之研究－以某農產品網路商店為例

Appling Data Mining Research on Consumer Electronic Commerce of Agricultural Products for an Agricultural E-store

曾羣峰(Tseng, Chunfeng)

指導教授 ： 彭克仲博士

國立屏東科技大學/管理學院/農企業管理系所/碩士(2012年)

https://doi.org/10.6346/NPUST.2012.00193

摘要

台灣電子商務市場規模逐年的成長，目前已成為消費者消費的主要通路之一，龐大的商機吸引許多產業業者投入經營，如何在如此競爭的市場中擁有競爭優勢，企業可藉由資料探勘（data mining）找出潛藏在龐雜資料中的有用資訊，例如發掘出消費行為背後的購物傾向，企業便可積極主動推薦貼切其需求的商品，以提升顧客的忠誠度及滿意度，並有效掌握顧客的心。
本研究使用Microsoft SQL Server 2008R2軟體，以決策樹及關聯分析兩項資料探勘技術，對某農產品網路商店會員資料庫進行探勘。
使用決策樹技術探勘發現，顧客購買商品類別主要影響因素為年齡
、收入及性別。如會購買加工農產品的顧客族群為，未滿55歲，且為女性的顧客，有63.38%的人會購買。個案公司可運用此特性，鎖定不同的影響因素，更精準的寄發E-DM給不同客戶群，以提高效率並節省成本。
關聯分析探勘出13個有用的關聯規則，如購買意麵與醬油膏的消費者，會100%再購買醬油。如此可將品項作網頁連結，或統一放置在專區網頁，提供消費者貼心服務，刺激顧客增加消費金額。
所以研究得知資料探勘可幫助企業，透過資料庫了解顧客之購買行為、找尋正確的顧客、擬定適合銷售策略，以達樽節行銷成本，創造最大盈利之目的。

關鍵字

資料探勘 ； 關聯分析 ； 決策樹 ； 電子商務

並列摘要

The e-commerce market in Taiwan is growing year by year, and becoming one of the main access for consumers. Such huge business opportunities attracts many industrial companies to put into operation. So it is very important to have a competitive advantage in such a competitive market. Companies can identify the useful information hidden in the numerous and jumbled data through data mining. For example, they can explore the shopping tendency hidden in consumer behaviors. As a result, the enterprises can recommend the appropriate goods in a proactive attitude to meet the consumer needs. They can enhance the loyalty and satisfaction of the customers, and grasp the hearts of customers effectively.
Two data mining techniques, decision tree and Association Rule Analysis of Microsoft SQL Server 2008R2 software, were used to the members’ database of an online store of agricultural products, and explored three cluster members.
After using the Decision Tree Analysis, we can find that the main factors influencing the consumers on buying the commodity categories were age,income, and gender. For example, Customer populations such as processing of agricultural products would purchase for less than 55 years of age, and female customers, 63.38% of people will buy. The case company can use these features and lock the different influencing factors, to send the E-DM to improve the efficiency and save the cost.
The Association Rule Analysis had mined 13 useful association rules. For example, consumers who purchased pasta and soy sauce will be 100% to buy soy sauce. So we can link the web links between the relational commodities items, or unify placed in the pages of the specified zone. Thus we can provid consumers with intimate services, to stimulate the customers to increase the amount of consumption.
According to the study, data mining can help companies to understand customers’buying behaviors, find the right customers, and to develop suitable marketing strategies to save the marketing costs to create the largest profit purposes.

並列關鍵字

Data Mining ； Association Rule Analysis ； Decision Tree Analysis ； E-Commerce

參考文獻

許嘉需，2007，應用資料探勘技術於人身保險新商品開發之研究，

陳彥良、凌俊青、許秉瑜，2001，在包裹式資料庫中挖掘數量關聯規則，資訊管理學報，7卷，2期，215-229。

梁定澎，2000，電子商務理論與實務，台北：華泰。

王貳瑞，2000，電子商務概論，台北：華泰書局。

陳垂呈，2007，利用分類分析發掘消費者最適性之產品項目，輔仁管理評論，15卷，1期，17-42。