不同世代及價格帶間 手機屬性差異之研究;Analysis of Cellphone Attributes in Different Eras and Price Zone

NCU Institutional Repository > 管理學院 > 企業管理研究所 > 博碩士論文 > Item 987654321/85885

jsp.display-item.identifier=請使用永久網址來引用或連結此文件: http://ir.lib.ncu.edu.tw/handle/987654321/85885

题名:	不同世代及價格帶間手機屬性差異之研究;Analysis of Cellphone Attributes in Different Eras and Price Zone
作者:	趙韻涵;Chao, Yun-Han
贡献者:	企業管理學系
关键词:	消費者偏好;網路評論;價格帶;Latent Dirichlet Allocation (LDA);情感分析;Ensemble Neural Network (ENNM);Customer preference, Online reviews, Price zone, Latent Dirichlet Allocation (LDA), Sentiment analysis, Ensemble Neural Network (ENNM)
日期:	2021-07-21
上传时间:	2021-12-07 11:37:49 (UTC+8)
出版者:	國立中央大學
摘要:	電子商務已成為人們生活中的一部分，進行線上購物的人越來越多。由於網路購物風險較實體購物高，許多人在線上購物前會搜尋產品在網路上的評論，之後才做出消費決策。因此，網路評論對於消費者而言越來越重要。此外，網路評論亦可以幫助企業了解消費者偏好，預測消費者滿意度。手機是會隨著科技而一起進步的產品，本文假設消費者對於手機屬性的偏好在不同世代及價格帶是會改變的，目的是為了找出在不同世代及不同價格帶上，消費者所重視的手機屬性是否有改變，並找出各世代及各價格帶上，消費者購買手機的趨勢為何，以提供手機廠商研發新手機時參考。本篇研究首先將評論分為三個時間帶，再將每個時間帶內的評論分為低、中、高三個價格帶，之後將分出來的 9 群資料(3 個時間帶3 個價格帶)分別使用 LDA 找出屬性，透過情感分析找出客戶提到屬性時的情緒分數，最後使用 ENNM 找出各屬性在神經網路的權重，以計算出消費者所注重的屬性重要度，最後將屬性依重要度排名。;E-commerce has become an undivided part of our daily life. More and more people use online shopping to buy products they want. Since online shopping is much riskier than physical store shopping, lots of people would browse product reviews before shopping online, and then make decisions. Therefore, online review has become much more important to consumers. Nonetheless, online reviews could also help companies understand consumer preferences and predict consumer satisfaction. Mobile phones are product that would evolved through technological progress. This article assumes that consumers preference for cellphones will change in different eras and price zones. The purpose of this research is to find out whether consumers preference would differ from ages and price zones, figure out the pattern of customers purchase behavior toward mobile phone in different ages and price zones, and provide cellphone manufacturers advices to develop new products. This research first divides online review data into three eras: touch screen, touch ID (fingerprint scanner), and face ID, and then divides data in each time zone into three price zones: low, medium, and high. After dividing data into nine groups (3 time zones 3 price zones), the proposed method has been executed. LDA is used to extract attributes from each data groups. Next, sentiment analysis is used to obtain the sentiment score when customer mentioned the extracted attributes in reviews. In the end, ENNM is used to find the weight of each attribute in the neural network to calculate each attribute’s importance. Afterwards, rank attributes according to the calculated attributes importance.
显示于类别:	[企業管理研究所] 博碩士論文

文件中的档案:

档案	描述	大小	格式	浏览次数
index.html		0Kb	HTML	105	检视/开启

在NCUIR中所有的数据项都受到原著作权保护.

社群 sharing

数据加载中.....