碩士 / 開南大學 / 商學院碩士在職專班 / 107 / In view of the rise of e-commerce and new retail, how to find their target customers and launch service innovation becomes the important goal to most of the companies. This study takes the example of an underwear company’s e-commerce website. First uses the data collected by Google Analytics to analyze the users of the website. The second part is to use the user data of at least one working phase starting from October 1, 2018 to December 1, 2018 as training data, and input into the logistic regression model to predict whether the user will purchase. The result shows that logistic regression can effectively predict the users purchase behavior, with 98% accuracy and about 50% recall, which means that about half of the users purchase behavior can be predicted by the model. According to the regression coefficient of the logistic regression model training, it is found that among all the user characteristics, the "Session" has the highest odds ratio, which is the most important feature for discriminating whether a user purchases or not. Speaking to device, the most important feature is "tablet", and the "mobile phone" used by most users is the last one. Therefore, A Company can try to improve the APP optimization.
| Identifer | oai:union.ndltd.org:TW/107KNU01318007 |
| Date | January 2019 |
| Creators | LIN,CHI-LAN, 林己嵐 |
| Contributors | LEE,ZONG-YAO, 李宗耀 |
| Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
| Language | zh-TW |
| Detected Language | English |
| Type | 學位論文 ; thesis |
| Format | 49 |
Page generated in 0.0091 seconds