| 摘要: | 線上拍賣是近年來相當盛行的C2C 網路交易方式，由於進入市場的門檻低，造成相當多的賣家競相投入創業。對於拍賣新手而言，常常因為不熟悉市場資訊而造成經營上的損失；也有些賣家只憑直覺經營，難以控制成本而且風險很高；又或是以試驗多種銷售策略以找到好的方式，造成不必要的損失，如此方法，皆不利長久經營。目前賣家的解決方法，有些是自行查詢歷史拍賣網頁，也有些是購買市場調查公司的服務，或是模仿他人的銷售策略。自行查詢歷史拍賣網頁，必須花費大量的時間，自行分析未整理過的資訊；市調公司的資訊常是僅限於單一因素對結標價和售出機率的統計資料；而模仿他人的銷售策略，則不見得適用，因為同一銷售策略的拍賣結果隨賣家及商品內容而有所不同。有鑑於此，這個計畫收集了eBay 拍賣網站的拍賣資料，預測商品的售出機率及結標價，進而推論出賣家的期望獲利，作為賣家銷售商品的參考。除此之外，我們也透過關聯規則探勘來找出多樣的銷售策略，尤其是高獲利的銷售策略，以滿足不同賣家的經營需求。 ; AsWeb become part of our lives, online auction has become one of the most popular methods for customer to customer transaction. Since the low threshold and cost to start a virtual store online, there have been a lot of virtual stores on auction web sites, not just second-hand products. For many novice users, they have to pay certain “tuition”to learn how to run their business. Some sellers run the business by intuition, thus, there is no way for them to control the cost and risk. For others, try and error to find a good selling strategy is also costly. To find a better way to run their business on these auction Web sites, some will survey auction history, some will buy statistics like business report, and some simply mimic others. For the first method, it would take a lot of time to analyze the data and there’s no standard way how to make decision. For the second, such statistics from business report are usually general, not specific to the scenarios the seller desire. As for the third, others’selling strategy might not work unless they are in the same business. In this project, we propose the use of cost-sensitive learning to model the problem of predicting the profit of a product. We will use the data from ebay to predict the probability of selling a product as well as the prices to be sold. Based on these two values, we should be able to compute the expected profit for a product using cost sensitive learning. After the prediction model is build, we try to mine the best selling strategy by dividing the data into 3 groups: high-profit, average-profit, and low-profit. The prediction model built is then used to verify if the strategies used for each group really affect the profit gained. ; 研究期間 9708 ~ 9807 |
