Empirical Research On Intelligent Prediction Of Financial Risk Of Enterprises

Enterprise financial analysis plays an extremely important role in the modern enterprise management decisions. Enterprise’s financial position is the lifeblood of organizations. Financial risk analysis can not only comprehensively reflect the operating results of an enterprise; also can provide reliable information for the majority of investors and stakeholders in decision making support. With artificial intelligence in the computer field infiltration and the continuing development of the artificial intelligence technology, the intelligent decision supporting systems of financial risk analysis will be continuously improved. An effective forewarning system of financial crisis can help listed corporations to forecast financial risk early and reduce loss. However, the financial risk analysis research often builds in assumptions that the distribution of data sets is balanced. The financial risk analysis research based on imbalanced data sets should not be ignored and still remains to further study.This paper adopts the method of interdisciplinary and theory analysis to research the financial risk in Chinese Tourism and hotel industry, including default risk, financial analysis of the indicators, data mining, forecasting and decision making, computer technology, multiple classifiers integrated technology, the financial risk forewarning and statistical sampling.Several relevant works have been done on the financial risk of Chinese Tourism and hotel industry under the conditions of non-balance and limited knowledge.First of all, in order to solve the problem of unsatisfactory results of unbalanced risk prediction on minority class samples, we suggested to adjust the up-sampling approach to be the neighborhood triangular synthetic minority oversampling technique (NT-SMOTE). The new approach that we add the nearest neighbor idea and the triangular area sampling idea to the SMOTE performed better in dealing with samples of minority class by turning imbalanced problems into balanced ones. Thus, performance of single classifiers in predicting risk on imbalanced and small datasets was improved. By using the related knowledge of data excavation principles, the data of listed companies of the Chinese tourism and hospitality industry were processed. Missing samples and missing financial indicators were eliminated. Significant indicators of financial data were filtered out with significance test. Then, NT-SMOTE was used to oversample minority samples. Further, we used a variety of popular single classifiers of financial risk prediction, including:multiple discriminant analysis(MDA), logistic regression(Logit), probit regression(Probit), decision trees(DT), linear support vector machine(LSVM) and Multi-classifier fusion(MCF), for risk prediction. These single classifiers improved with NT-SMOTE can reasonably and effectively solve the problem of imbalanced and small sample oriented firm risk prediction.The second, in order to improve effect and stability of classifiers, the paper builds six new combiners by using the bagging algorithm and achieves a high performance to some extent, including BMDA, BLogit, BProbit, BDT, BLSVM and BMCF.And then, in order to better assess the credit risk of tourism and hospitality firms under the condition of limited knowledge and information, this article presents the heterogeneous data mining case-based reasoning(HDM-CBR), including data preprocessing, neighbors triangle increment, extraction heterogeneous neighbors, case-based reasoning. In the demonstration research, we collected data about Tourism Hospitality Firms, excluded missing data by preprocessing, obtained significant financial indicators using statistical tests, increased minority class cases using neighbors triangle increment, built early warning models using HDM-CBR, including HDMMDA, HDMLogit, HDMProbit, HDMDT, HDMLSVM, HDMMCF, HDMBMDA, HDMBLogit, HDMBProbit, HDMBDT, HDMBLSVM and HDMBMCF. Empirical evidence shows that the new method is signficiantly better than previous methods in terms of credit risk prediction.Finally, in order to effectively improve and enhance the traditional risk prediction methods of tourism and hospitality firms under the limited knowledge and unbalanced environment, the article presents multiple case-based reasoning models based on the neighborhood case reuse and their fusion technology(CBR-R) including the layer of data preprocessing, the layer of triangle neighborhood incremental,the layer of unbalanced neighbors extraction, and the layer of case reuse prediction. In the demonstration research, we collected financial data about the chinese tourism and hospitality firms, excluded missing data by preprocessing, obtained significant financial indicators using statistical tests, increased minority class cases using neighbors triangle increment, built early warning models using CBR-R, including RMDA, RDT, RLSVM, RLogit, RProbit, RMCF, RBMDA, RBDT, RBLSVM, RBLogit, RBProbit and RBMCF. Empirical evidence shows that the new method is signficiantly better than previous methods in terms of credit risk prediction of the Chinese Tourism and Hospitality Listed Firms.