为了了解网商的典型发展模式和影响网商发展模式的重要因素,与以往众多的定性研究不同,提出了一种采用隐马尔科夫模型定量分析网商发展模式的方法。在此基础上通过不等长的时间序列聚类技术展示行业内网商主流的发展模式,并采用多分类logistic回归的方法分析与发展模式相关的重要因素。最后,以淘宝社区1092个母婴类产品卖家的销售数据为例验证了方法的可行性。
To understand growth pattern, as well as the critical factors about the growth pattern, this paper presents a quantitative method which applies the Hidden Markov Model (HMM) and clus- tering of time series data to obtain the primary pattern of growing process. Meanwhile, multinomial regression is also applied to further analyze the factors relevant to the growth pattern. The experiment on the 1092 Taobao entrepreneurs is provided to demonstrate the feasibility of the method.