Financial Early Warning Of Internet Listed Companies Based On Random Forest

In 2018,the Internet industry entered the fast lane.The new Internet economy is widely developing in China,and has become a strong driving force for China's transformation and upgrading.Internet companies have ushered in the third listing boom this year,and many emerging Internet companies have begun to enter the international arena.However,while the Internet industry is improving,it cannot ignore the financial crisis that always occurs.The common problems in the industry,such as debt-raising operation,rapid expansion,unreasonable capital structure and single investment mode,have also put many companies in a dilemma.In 2018,85%of Internet companies' market capitalization showed a negative growth,the entrepreneurial heat cooled,and the phenomenon of acquisition collapse occuired frequently.Therefore,considering the promotion effect of Internet enterprises on economic transformation and unique financial attributes,it has certain practical significance for Internet companies to identify and warn financial risks.The traditional financial early warning model has low matching with the financial data of the Internet industry.In the past,the criteria for determining whether the financial situation of listed companies is abnormal could not effectively classify Internet enterprises.The time limit of early warning is too long for the rapidly changing Internet enterprises,and the traditional early warning model is poorly explained.In response to these problems,the paper has carried out the following work:Firstly,it summarized the previous research,constructed a comprehensive financial indicator system,and introduced the random forest theory.Secondly,using the factor analysis to calculate the comprehensive factor score,realize reasonable division of Internet enterprise's financial capacity.Finally,using the random forest construction early warning model,the model is optimized to solve the unbalanced classification problem,the optimal model is obtained by comparing the different models,and the importance of financial indicators is exploined.The final conclusion is that the random forest model which solves the unbalanced classification problem by voting weight method has the lowest cost and the best prediction effect Net profit growth rate,cash flow ratio,return rate of total assets,operating leverage coefficient and other indicators are highly important for judging the financial ability of the Internet industry.So enterprises should pay more attention to the corresponding financial status in daily management.