Research On Financial Crisis Early Warning Of Real Estate Listed Companies Based On PCA-SVM Model

In the context of economic globalization,more and more enterprises attach importance to financial crisis.If an enterprise can predict its own financial crisis in advance,it can take precautions to avoid losses to itself and other stakeholders.In order to predict their own financial crisis,enterprises are required to have an effective financial crisis warning model to evaluate whether financial crisis will occur.Since the 21 st century,China’s economy has developed at a high speed,and the real estate industry has contributed indelible strength to it.But now China has turned the pursuit of high-speed economic development into the pursuit of high-quality economic development,and has subsequently introduced many measures for the real estate industry to carry out macro control.At the same time,the new crown epidemic has also brought a huge impact on the overall economic environment in recent years,and the real estate industry can be said to be in a situation of internal and external troubles.The real estate industry itself is a highly indebted industry,so it is more important to carry out financial crisis warning for it.For the real estate enterprises whose financial situation has deteriorated,if they cannot judge their financial situation as soon as possible,predict their financial crisis in advance,and make them improve their financial ability or governance level,then they are likely to go bankrupt.Therefore,the research on financial crisis early warning model of real estate listed companies has extremely important practical significance.This paper includes the following two parts of research content.The first part is the construction of the financial crisis early warning indicator system.First,on the basis of the research results and theories of domestic and foreign scholars,the qualitative research method is used to select eight categories,including solvency,operating capacity,profitability,development capacity,cash flow indicators,EVA indicators,per share indicators and non-financial indicators,totaling 31 indicators.After that,the quantitative research method was used to conduct a significance test on the selected indicators,and 19 indicators with significance were retained.Due to the problem of collinearity among indicators,PCA principal component analysis was selected to reduce their dimensions,and six principal component factors were obtained on this basis.The second part is to use SVM support vector machine model to predict the financial crisis.This paper selects 71 sample enterprises from 2017 to 2021,selects their financial data for T-3 years,and randomly groups them.Among them,80% of the samples are training groups.After learning through SVM,the remaining 20% of the samples are predicted.The accuracy of the model is 81.25%,of which the error rate of the first type is18.75%,and the error rate of the second type is 0%.Finally,according to the early warning model,the paper puts forward targeted suggestions for financial crisis early warning of real estate listed companies,which has a certain guiding role.