| DC 欄位 |
值 |
語言 |
| DC.contributor | 企業管理學系 | zh_TW |
| DC.creator | 鄒雅淳 | zh_TW |
| DC.creator | Ya-Chun Tsou | en_US |
| dc.date.accessioned | 2019-7-9T07:39:07Z | |
| dc.date.available | 2019-7-9T07:39:07Z | |
| dc.date.issued | 2019 | |
| dc.identifier.uri | http://ir.lib.ncu.edu.tw:88/thesis/view_etd.asp?URN=106421046 | |
| dc.contributor.department | 企業管理學系 | zh_TW |
| DC.description | 國立中央大學 | zh_TW |
| DC.description | National Central University | en_US |
| dc.description.abstract | 隨著國人健康意識的抬頭，保健食品市場規模呈現逐年上升，使得企業致力於發展多元的行銷管道，其中又以線上銷售的成長最為快速。而電子商務的蓬勃發展以及消費者消費行為的異質化使得個人化的推薦系統越來越受到重視。然而，這些推薦系統卻不適用於電話行銷產業，原因包含：(1)過去的推薦系統習慣對所有消費者進行商品推薦，但在電話行銷產業中，此舉會消耗大量的人力及成本；(2)協同過濾推薦系統所使用的評分機制對電話行銷來說有一定困難度；(3)傳統推薦系統使用的信任評量方法在電話行銷中難以實現。因此本文從ERP系統中萃取資料，利用卷積神經網路為演算，以信任以及忠誠度先行預測客戶的再購行為，進行客戶篩選，再以TF-IDF為概念所計算之商品分數向這群客戶進行推薦，這項做法有助於減少電話行銷人員之撥打客戶數、減少通話成本。實驗證實本研究模型之轉換率明顯高於直接向所有客戶進行推銷之轉換率。 | zh_TW |
| dc.description.abstract | With the rise awareness of health, the scale of the nutritional supplement market has been increasing year by year. Making the companies devoted to developing diversified marketing channels. And online sales have grown the fastest among them.
The booming of e-commerce and the heterogeneity of consumer behavior have made personalized recommendation systems more and more important. However, these recommendation systems are not suitable for the telemarketing industry. The reasons are:
(1) Previous recommendation system recommended products to all customers. But in telemarketing, this will consume a lot of labor time and costs;
(2) The rating mechanism used by the collaborative filtering recommendation system is difficult to achieve in telemarketing;
(3) Tradition trust assessment method used by the recommendation system is not suitable in telemarketing.
Therefore, this study extract data from ERP system and apply the convolutional neural network as the algorithm, predict the customer′s repurchase behavior with trust and loyalty to filter out those who won’t purchase again. And then make recommendations to the customers who will repurchase. This will help to reduce the number of calls made by agents and cut the cost of calls. The experiment result shows that the conversion rate of our research model was significantly higher than the one which made calls to all customers.
| en_US |
| DC.subject | 推薦系統 | zh_TW |
| DC.subject | 再購行為 | zh_TW |
| DC.subject | 信任 | zh_TW |
| DC.subject | 忠誠度 | zh_TW |
| DC.subject | 卷積神經網路 | zh_TW |
| DC.subject | Recommendation system | en_US |
| DC.subject | Repurchase behavior | en_US |
| DC.subject | Trust | en_US |
| DC.subject | Loyalty | en_US |
| DC.subject | CNN | en_US |
| DC.title | 預測客戶再購與所購品項之研究 | zh_TW |
| dc.language.iso | zh-TW | zh-TW |
| DC.title | Prediction of customer repurchase and purchased items | en_US |
| DC.type | 博碩士論文 | zh_TW |
| DC.type | thesis | en_US |
| DC.publisher | National Central University | en_US |