A STUDY ON PREDICTING BROWSING AND PURCHASING ITEM USING EXPLORATORY BEHAVIOR AT EC SITE

Ko Hashimoto, Graduate School of Science and Engineering, Chuo University, Bunkyo-ku, Tokyo, Japan
Kohei Otake, Faculty of Science and Engineering, Chuo University, Bunkyo-ku, Tokyo, Japan
Takashi Namatame, Faculty of Science and Engineering, Chuo University, Bunkyo-ku, Tokyo, Japan

Published in

JOURNAL OF INTERNATIONAL MANAGEMENT STUDIES
Volume 17, Issue 3, p41-48, December 2017

ABSTRACT

With the spread of the Internet, online shopping users have increased. It is very important and general to predict future consumer behavior because EC sites utilize on services such as recommendation and Real-Time Bidding (RTB). Conventionally, commonly used models of consumer behavior are machine learning such as neural networks and random forests. However, these methods are used by converting behavior histories into fixed-length vectors and the features of time series data are not captured correctly. In this study, we estimate the utility with respect to items from the consumer exploratory behavior in real time and predict item category that the consumer will purchase. Especially, we analyze access log data of consumer behavior provided by the golf portal site in Japan. We use Recurrent Neural Networks (RNN) to learn a series of consumer exploratory behaviors. We show that LSTM is superior to conventional machine learning method from the view point of prediction accuracy and it is useful as a method for grasping utility in real time. Moreover, it is very effective for consumers who continuously show high utility.

Keywords

Electronic Marketing, Access Log Data, Consumer Behavior, Recurrent Neural Networks


About the Article

Abstract, Keywords, Page Numbers, etc

About the Journal

Managing Editors, Indexing, Best Practices

About The Publisher

History, Partners, Conferences

Access the Full Article

Log-in to IABE to access full article

Search IABE

Search IABE's articles by Title, Author, or keyword

Contact Us

Send a message to IABE