摘要
信用问题是电子商务发展中的重要问题,也是用户最关注的问题之一。笔者在分析我国企业信用评估方法的基础上,指出了网络供应商信用评估中存在的问题;以B2C电子商务为例,提出网络供应商信用评估数据的来源及数据处理流程,分析了基于大数据的网络供应商信用博弈行为,使用逐步回归法建立了B2C网络供应商的信用评估指标体系,将逐步回归法的指标体系线性回归参数值作为指标变量的权重,构建基于部分调整模型及带权重的径向基神经网络模型的B2C供应商信用评估模型。通过实证分析并与纯线性回归模型、纯径向基神经网络模型的信用评估结果进行比较说明本研究模型的有效性。在考虑各期信用水平滞后影响关系的前提下研究B2C供应商当期的总体信用水平,将网络供应商放到大数据环境中构建信用评估指标体系,综合计量经济学方法与数据挖掘算法构建信用评估模型,具有一定的前瞻性和很高的实用性,对于提高网络供应商信用评估水平具有一定的参考价值。
The credit is the important issue in e-commerce＇s development,and is the most concerned problem of customers. Based on enterprise＇s credit evaluation method,we point out the shortcomings of network suppliers＇ credit evaluation. Taking B2 C e-commerce as an example,we propose network suppliers＇ credit evaluation data sources and data process flow,analyze the network suppliers＇ credit game behavior based on big data,construct B2 C network suppliers＇ credit evaluation index system by using stepwise regression method,take the linear regression parameter values of the stepwise regression method index system as the weight of index variable,construct B2 C supplier credit evaluation model based on partial adjustment model and weighted RBF neural network model. We run the empirical analysis and compare it with pure linear regression model,pure RBF neural network model to illustrate the research model is effective. Considering the influence of each period＇s credit level lag,we study B2 C supplier＇s total credit level of current period. We set network supplier into big data environment to construct credit evaluation index system; we also integrate econometrics methods and data mining algorithms to construct credit evaluation model. The model is practical and quite usable in improving network supplier＇s credit evaluation level.
出处
《中央财经大学学报》
CSSCI
北大核心
2016年第8期74-83,共10页
Journal of Central University of Finance & Economics
基金
国家自然科学基金项目“基于博弈论的信息安全理论与方法研究”(项目编号:61272398)
国家社会科学基金重点项目“大数据时代网络媒介生态环境下个人信息保护体系的构建研究”(项目编号:13AXW010)
北京市哲学社会科学重点项目“基于大数据的财务预警理论与方法研究(项目编号:14JGA001)
