The Financial Crisis Warning Research Of The Listed Companies In China

With the continuous development of the securities market in our country, the government, investors and managers’ concerning for operating performance of listed companies are also gradually increasing. If the listed companies fall financial crisis in the future, not only its own survival and development will be impeded, but also it can make investors suffer losses, and even may affect the stability of the securities market. Accurate early warning for financial crisis can protect the interests of investors and creditors, help operators prevent financial crisis, and be able for market regulators to track the operating performance of listed companies and the securities market risk, which has very important practical significance.There are a lot of methods used to study the financial crisis warning problem at home and abroad, this article selects the latest neural network model and support vector machine model, and comprehensively contrasts the theoretical and empirical results between the two models. First of all, on the basis of existing research, this paper defines the financial crisis, characteristics and causes of the listed companies in our country, and instroduces the selection method of indexes system for financial crisis warning. Than it respectively introduces the theories of neural network and support vector machine. After analyzing the theories of the two models, the author thinks that the two models are different in theoretical basis, model structure, algorithm and development degree. Next, through principal component analysis, the paper extracts principal component of the indexes, respectively ueses the neural network model and support vector machine model, selects ST listed companies and non-ST companies in our country as samples to do the empirical analysis of company financial crisis early warning. Finally, According to the research results, the paper rises policy recommendations for the supervision and administration institution investment institution.Through comparison of the two models’ theories, the paper finds that the neural network model is suitable for large sample analysis, and support vector machine model is more effective in small sample study. A neural network model with good effect requires the designer’s certain experience and prior knowledge, while support vector machine doesn’t need such requirement. The neural network model can solve multiple classification problems, while support vector machine is used to solve binary classification problems. On the other way, after a comprehensive comparison between the results of prediction, the author finds that when the modeling sample size is large, the neural network model and support vector machine model are suitable for listed company financial crisis early warning in our country, while the effect of support vector machine model is slightly better; When modeling sample size is rare, support vector machine model is significantly better than neural network model for financial warning, support vector machine can still apply to the financial crisis early warning, while neural network model is no longer applicable.