| At present,under the high attention of the country,the overall development of high-tech enterprises has advanced by leaps and bounds.In 2016,the management measures for the identification of high-tech enterprises were revised again;in 2019,the Science and Technology Board was established to promote the deep integration of technological innovation and capital markets.So far,on the issue of how to increase support for scientific and technological innovation,China’s capital market has carried out a lot of exploration and effortsAfter the outbreak of the global financial crisis,all countries paid special attention to the financial risks and financial crisis of enterprises.How enterprises can early warning the financial crisis and take corresponding effective measures to enable the enterprise to effectively stop the financial crisis in its infancy stage is worthy of further discussion and investigation.At the same time,compared with ordinary companies,the high-tech enterprises’ huge investment in R&D expenditures and the high degree of uncertainty in the conversion rate of R&D innovation achievements will inevitably double the possibility of financial crisis.Therefore,in order to improve the ability of high-tech enterprises to deal with financial risks and defend against financial crises in the face of fierce market competition,and reduce the possibility of financial crises,so as to promote the steady development of enterprises,the relevant explorations for high-tech enterprises’ financial crisis warning important.This article combines the characteristics of high-tech enterprises,studies and analyzes the causes of financial crises and early warning processes,and uses this as a basis to select two types of comprehensive early warning indicators that combine financial information and non-financial information.The cross-sectional data of technology enterprises and non-ST(normal)high-tech enterprises in the past two years are used as research samples.Based on the characteristics of the selected indicators,the Kolmogorov-Smirnov test,the T test for independent samples,the Mann-Whitney U test and the Chi-square test are used to select early-warning indicators.The packaging method in feature engineering is used for further screening to establish a simple and effective index system for early-warning research on financial crisis of high-tech enterprisesChoose Logistic regression model,XGBoost model and BP neural network model as the base model,and use genetic algorithm to optimize the weights and thresholds of the original model of BP neural network,fuse the three by Stacking method,and train the fused model And forecast,and at the same time compare the fusion of different quantity-based models and the fusion methods of Voting and Averaging to build a high-tech enterprise financial crisis early warning model.In this process,in order to make the effect of each model more stable,the SMOTE principle in the upsampling method is used to process the unbalanced data set,and the 5-fold cross-validation method is used to select and adjust the model parameters to achieve the optimization of the stable model.purpose.The results show that compared with the common Voting and Averaging model fusion methods,the Stacking method fusion model is more robust and has better performance.Comparing the fusion effects of different quantity base models,it is found that the larger the difference between the base models and the closer the performance,the better the model fusion effect of the Stacking method.In terms of quantity,the fusion effect will not continue to be optimized.Finally,based on the modeling process and the inspection results,the preventive measures that high-tech enterprises can adopt in response to the financial crisis are targeted to ensure the vital interests of other stakeholders such as investors and creditors,and provide regulators with the management of high-tech enterprises.Certain references have strong theoretical value and practical significance. |
PDF Full Text Request