基于支持向量机的用户行为聚类分析

In order to supply better service for network customers, deeply analyze customers' behavior is required first. This paper extracts three types of characteristics from customers' network behavior and divided the behavior in to different categories, such as browsing news, download shared resources and network communications etc. This paper uses support vector machine to perform clustering, because this method is proved as fast and valid, especially in the situation of small datasets. After acquired the analyze results, we can make our application and services more personalized and easier to use.