基于神经网络的卷烟零售户销售假烟行为的预警模型研究

A neural network-based early warning model for counterfeit cigarette traded by retailers

在卷烟零售户销售假烟驱动因素分析的基础上,通过对安徽某市烟草专卖局的实地调研以及所获得历史数据的系统分析,构建了基于BP神经网络的零售户销售假烟行为的预警模型,该模型综合考虑了零售户的销售行为、零售户自身的特征以及时间因素等十二个预警输入指标。实证表明所构建的模型稳健性较好,预测准确率较高。

Field study and data analysis were conducted in a city of Anhui province to establish a early warning model of counterfeit cigarette sales based on BP neutral network by analyzing retailer＇s motivation for counterfeit selling. This model took into account twelve factors such as retailers selling practice, their individual characteristics and time factors. It was proved to have good stability and improved accuracy of early warning.