Purchasing Behavior in Electronic Commerce Network Based on Fourier Transform and DFA

Shuaili WANG

Hunan City University, Hunan, Yiyang 413000, CHINA

Abstract: In the electronic commerce network, the purchasing and browsing behaviors of the users have the statistical properties and scaling law of the time series. The empirical evidence finds that the probability dis- tribution of the time series of the purchasing and browsing behaviors have the significant dual modal charac- teristic, and its scaling behavior shows a strong long-range correlation characteristic. This paper uses the Fourier transform and the DFA method to analyze the user behavior of the network, through the spectral anal- ysis of the Fourier transform, this paper finds that the time series of the purchasing and browsing behaviors both have long-range positive correlation, and their main frequencies and the corresponding periodic trends can be found out and filtered according to the spectrum diagram. After filtering the main periodic trends, the DFA analysis can effectively obtain the true scaling behavior of the time series. The average scale value is approximate to 1, which shows that the evolutionary process of the online shopping activities has the self- organized criticality. The experimental results show that the method in this paper is correct and effective.

Keywords: Fourier transforms; User behavior; High frequency