Research On Financial Crisis Early Warning Algorithm Based On Random Forest And Time Series Analysis

With China's economic development entering a new normal and the government encouraging "mass innovation and entrepreneurship",the domestic economic environment is developing rapidly.Government regulators,professional financial institutions,corporate decision makers,and investors need to be able to keep abreast of the latest corporate data information,and expect to predict the further development direction of the company from the current financial data of the company,and make the next step in advance.This requires the establishment of a financial crisis early warning system that can be dynamically analyzed based on historical data.Aiming at the research goal,this thesis proposes an enterprise financial crisis early warning algorithm combining K-fold random forest algorithm and time series analysis model.The algorithm uses the time series analysis model to predict the historical data in a short-term and uses the time series analysis model to predict the newly constructed financial indicator data.Using the K-fold random forest to analyze the financial status of the forecasted financial data,the purpose of the dynamic financial crisis early warning is realized.In this research,this thesis uses the advantages of random forests that do not require excessive prior knowledge and good anti-noise ability and high classification accuracy to achieve the classification of corporate financial status.In the process of the financial crisis early warning,the sample set sampling method of classification and regression tree of traditional random forest algorithm is improved,and a K-fold random forest algorithm is proposed.The algorithm makes the feature attributes selected when the decision tree is split better than the original algorithm and improves the classification accuracy of the original algorithm.In addition,according to the feature importance degree algorithm,the financial indicators are sorted in importance,and the financial indicators are reduced in dimension,and a new financial indicator system is established.This thesis takes 92 listed companies in the manufacturing industry as the research object and analyzes the financial indicators in five aspects.The experimental results show that the classification accuracy of the K-Fold random forest algorithm is better than the traditional random forest algorithm,and the new financial indicator system can Better reflect the financial situation of the company.Finally,experiments show that the prediction accuracy of the financial crisis early warning model based on random forest algorithm and time series is 89%,which indicates that the model is effective and feasible.