- 學位論文
應用資料探勘挖堀電子商務潛在客戶群
using data mining techniques to dind out potential coustomers in e-commerce marketing
摘要
目前電子商務的市場經之前的泡沫化後又一波的熱起，現在的金流及物流都比以前更方便而且一般民眾都接受網路交易的模式，個各公司分別也提供電子商務的平台但是真正會使用電子商務或是潛在的客戶群在哪裡是值得我們探索的。 在這篇論文當中利用資料探勘技術來探討現行的潛在的客戶及那些族群是主動的客戶哪些族群是被激發的被動型客戶，我利用各家電子商務軟體使用的客戶作為訓練的樣本歸納出使用電子商務的共通性在把一些隱藏的會使用的因素用資料探勘找到潛在型客戶。資料探勘技術有許多不同的演算法，此次採用Association rule的Algorithm Apriori方式。Apriori 演算法最主要的目的是要找到高頻項目組，其定義為：符合最小支持度的所有項目組，若一個項目組含有k個項目稱之為k-項目組，若滿足最小支持度者，則稱為高頻k-項目組；找到高頻項目組之後，我們再以最小信賴度為條件，來判斷所形成的關聯規則是否成立。 第一步：分析現有客戶加入電子商務的因素如：使用者的年齡、開店商品、是否為實體店家、有無網路上販售商品經驗．．．等，並給選出真正影響決定使用電子商務平台因素再經訓練出演算規則後，在依每次輸入值在做調整，驗算出來的結果以百分比來代表客戶所要使用電子商務的渴望及需求，並分析為何會使用電子商務的原因，如:他是否之前為拍賣族群並有在網路販賣商品，所以對網路開店有相當的經驗及信心或是為企業目前的需求而選擇一家作為合作的資訊廠商．．．等，最後經過程式判斷的結果並漿現有的市場資料經程試解析後真正會使用或是成為電子商務的客戶的機率有多少，再由業務人員實際依程式的建議去推廣並紀錄實際的結果，旗本論文有經實際的演練，大約精確率達百分之四十左右有些誤差的原因可能是因為每個人的個性無法在程式內判斷或將此因素加入到判斷的條件內，導致實驗結果有誤差為重要原因之一，希望提供各大目前電子商務界公司作為分析潛在客戶族群者工具參考。 資料探勘:是經由自動或半自動的方法探勘及分析大量的資料，以建立有效的模型及規則在大量的資料中找尋可利用的資料;電子商務: 利用資訊網路進行的商務活動。
關鍵字
電子商務並列摘要
After the previous E-Commerce bubble has been broken a new era of e-commerce has just begun. Nowadays the majority of the public have high degree of acceptance in doing business over the Internet, each every company also provides E-Commerce platform, but where is the potential client in the market? It is worth to give us a deep thought. In this study, Data Mining approach will be used to discuss which group of client is more initiative or which one tends to be more passive, and to find out their identical characteristics and discrepancy. Some of the E-Commerce soft wares that are currently used by many users are my test samples. Thus, if the similarities characteristics of these soft wares have been found, data Mining approach is very helpful to find out the most potential e-commerce software user in the market. Data Mining has many different algorithms. Algorithm Apriori in Association rule is the method that we use in this study. The main purpose of Apriori is to find out the high frequency group, its definition: if one group meets the minimum level of support, and it contains k-value we call it as k-group, and if it satisfies the minimum level of support we call it as the high frequency k-group; after we found the high-frequency group, we use the minimum level of confidence to judge whether the rule of relativity is established or not. The first part takes the reason why the existing customers want to join e-commerce into consideration: the age of opening the store and whether they have former experience of selling product over the Internet, etc. and we input the information to obtain the result and presented it in percentage to show the reasons why the customer choose to use e-commerce and what are their demands. The result might show that having a great confidence on doing business online is because they have been successfully done it before, or it is because their demands for the company as they choose an information provider to be the working partner or etc. In the second part all data processing procedure interfaces will be shown vividly to have a better visual effect. In the third part, all data will be analyzed and current marketing data will be presented to the sales representative for them to practice in the real market and keep the record of the result. The margin error will probably around 40% to 50 % due to the unexpected individual behavior can not be taken into account during the calculation. Further explanation and solution on how to correct and suggest a plan is a big discovery and that help enterprise find many potential e-business customers.