基于多层前馈神经网络的互联网行业财务风险预警研究

为了构建一套科学合理、行之有效的财务风险预警机制，及时识别出企业隐藏的财务风险。本文以60家在A股上市的互联网企业为依据，以各企业2016~2018年的数据作为研究样本，选取了包含财务和非财务指标共计27项，构建了初步预警指标体系。通过非参数检验、主因子分析对指标进行优化和降维，将因子得分作为输入变量代入多层前馈神经网络进行训练和检验。检验结果表明：基于多层前馈神经网络的互联网行业财务风险预警模型达到了设计标准，具有较高的准确度和实际应用价值。

Research on Financial Risk Early Warning of Internet Industry Based on Multilayer Feedforward Neural Network

In order to build a scientific, reasonable and effective financial risk early warning mechanism, identify the hidden financial risks in time. Based on the data of 60 Internet companies listed in A-share market from 2016 to 2018, this paper selects 27 financial and non-financial indicators and constructs a preliminary early warning indicator system. The index is optimized and dimensionally reduced by nonparametric test and principal factor analysis, and the factor score is used as input variable to be trained and tested by multilayer feedforward neural network. The test results show that the financial risk early-warning model of Internet industry based on multilayer feedforward neural network meets the design standard, and has high accuracy and practical application value.