| Abstract: | 三維列印機(3d printer)打破了常規的供應鏈體系，也加速了產業智能轉型的腳步，隨著政府政策的支持及技術創新，全球市場也在持續成長，且隨著智慧型手機和物聯網的興起，企業更容易取得用戶資料，並渴望使用數據挖掘技術了解客戶。由於任職於三維列印機公司，本研究以實務應用為主，研究主題為預測消費者購買行為(customer purchase behavior prediction)，並探討三維列印機產品屬性對購買決策的影響。
本研究以 6,926 位消費者，共6,979 筆三維列印機訂單資料做為研究樣本，以聯合分析(conjoint analysis)編碼產品屬性，採用邏吉斯迴歸(logistic regression analysis)進行假設驗證。研究結果發現，經過總體消費者偏好分析後，相較於沒有任何用戶資訊輔助下，其產品推薦擊中率有正向成長，此外將消費者依照偏好分群後，推薦擊中率會更高。
A recent survey suggests that global e-commerce retail sales will reach 4.9 trillion US dollars in 2021, and the key to affecting e-commerce revenue is consumer consumption data. Analyzing consumer preferences and predicting purchase behavior is the key to the success or failure of e-commerce. However, our current understanding of this issue is still very scarce. By querying the customer's past historical purchase records in the database, combined with the conjoint analysis method, since the purchased goods are equivalent to the customer's subjective purchase preferences, subdivided the attribute levels of the products purchased by the customers, and quantify the users buy choices.
Customer relationship management must be continuous to find the most valuable customers, build a customer recommendation system, increase customer loyalty, improve the consumer experience, conduct precise marketing, and increase corporate revenue. This research will use the conjoint analysis method to analyze the best product attributes and purchasing preferences of the 3D printers on the market today in consumers' minds to provide developers with future R&D directions.
After interviews with industry technical experts in 3D printers, define 13 product attributes. Through cross-border e-commerce company, this study collected 28 products in 20 countries on four continents and a total of more than 6,000 customer database transaction data from 2014 to 2020.In terms of contribution, this research is the first time to analyze 3D printer customer preferences, conduct personalized marketing, and establish a customized customer preference prediction model. Second, this research integrates multiple product category data. The conclusion can expand cross-border e-commerce market analysis, market segmentation, and marketing combination strategies for 3D printers and more closely with customers. In practice, Taiwan's small and medium-sized company can, base on this conclusion, operate cross-border e-commerce, quantify user preferences, and develop the best marketing strategy exclusive to the enterprise. |
