Research On Financial Crisis Prediction Of Internet Corporations With PCA-SVM Model

By June 2018,the number of Chinese netizens had reached 802 million,and its penetration rate had reached 57.5%.China has the largest network of netizens in the world,and the strategy of building a digital power country has gradually been implemented,which has become the fertile soil for Internet companies to flourish.Internet companies have made great progress in platformization,dataization,generalization,and ecologicalization.However,due to the “big wild” era of Internet in the past decade,Internet companies have accumulated a lot of financial risks while growing rapidly.But there are few studies focusing on the sub-sectors like China's Internet industry,this phenomenon makes the establishment of a set a complete indicator with timely and effective and targeted Internet enterprise financial risk early warning model is imminent.Financial risk warning is a method that most intuitively reflects the current financial status of the current and future enterprises.It enables decision makers to quickly locate problems and difficulties in production and management,formulate effective countermeasures,and reduce the investment risk of investors.In terms of government regulation,it is possible to formulate a more accurate and effective supervision plan based on the early warning model.Therefore,strengthening the financial risk management of Internet companies has strong practical significance both at the macro level and at the micro level.The financial activities of listed companies on the Internet will be affected by the financing process,traffic economy,profitability,technological innovation,etc.Based on these influencing factors,preliminary screening of financial early warning indicators was obtained with 19 financial indicators and 12 non-financial indicators such as corporate governance capabilities and technological innovation capabilities.This paper firstly develops a rich theoretical foundation for the later research by combing the domestic and foreign literature on financial risk definition,financial early warning methods and Internet enterprise risk management.Furthermore,by identifying and summarizing the characteristics of China's Internet industry,the characteristics of Internet financial risks and the causes of risks.We try to describe the impact path of financial risks on enterprises from the industry,explore the financial risk early warning mechanism,and provide reference for selecting financial early warning indicators.After that we collected 257 sample data of listed companies in the Internet industry in China from 2015 to 2017.And 184 Internet-listed companies were selected as research objects.The Lasso regression is used to eliminate some low-correlation indicators.Subsequently,used SPSS 22 software to standardize the processing and factor analysis of the indicator data,and then through principal component analysis to reduce the dimension to obtain 7 principal components that can effectively reflect the financial status of the enterprise.Finally,used the Python language to establish PCA-SVM model data for empirical analysis.Compared with the Logistic model,BP neural network model for predictive rate analysis,the PCA-SVM model has higher prediction accuracy than the other two methods,and lower error rate,further confirm the PCA-SVM model The effectiveness and accuracy of the Internet enterprise financial early warning system.According to the empirical results,the countermeasures and suggestions for the financial risk control of the Internet industry are obtained.