Research on the Changes of Online Shopping Reviews based on Time Series Analysis
Download as PDF
Xiaoyi Xu, Yue Jiang
As e-shopping becomes one of the most successful innovation in the digital era, the on- line reputation of product occupies a more important role in the company’s mind. In this paper, we used the provided dataset from the Sunshine Company to discover patterns of comments in each of the three categories of product, helping the Company to enter online market with its brand-new inventions. To study comments based on time-varying patterns, we established the time series models using Rescaled Range Analysis method and Detrended Cross-Correlation Analysis method. Particularly, the methods we used overcome the limitations of ordinary time series models which cannot accurately analyze non-stationary time series. Based on fractal system, our time series model researched the auto-correlation, persistence and cross-correlation of the time series. The use of Hurst exponent and DCCA method could be the biggest innovation of this article.
Online Shopping Reviews, Time Series Analysis, Detrended Cross-Correlation Analysis, Rescaled Range Analysis