基于SOM的手机电子商务交易聚类分析

Clustering Analysis of E-commerce Transactions with Self-Organizing Map

针对电子商务研究中未充分重视卖方特点的研究现状,利用自组织映射(SOM)神经网络方法,对手机电子商务交易状况进行聚类分析,研究电子商务卖方的自身特点与其销售商品之间的匹配程度对交易活跃程度的影响,有利于电子商务卖方及产品生产商了解市场状况及自身优劣,以便改进其生产及经营活动。现有的SOM统一距离矩阵(U-matrix)的算法基础上进行修改,提出新的U-matrix算法与显示方法,并应用于本文的数据分析。

To improve the current situation that research on sellers＇ attributes has not received enough attention in the field of E - commerce study, this paper analyzes the E - commerce transaction data of cellular phones with Self - Organi- zing Map （SOM） and explores how the match between sellers＇ attributes and the commodities affect the prosperity of the transactions. The findings will help sellers and manufacturers understand the market status and their own advantages and disadvantages, and they can take measures to improve their production and operation. In the aspect of research methods, the algorithm of U - matrix has been modified based on the existing version, then a new definition and display of U - matrix are proposed and applied to the data analysis in this paper.