Research On Enterprise Financial Distress Prediction Based On Adaboost Algorithm Combined With DEGWO-SVM

Under the environment of the rapid development of economic globalization and the continuous escalation of the Sino-US trade war,how can our country's listed companies maintain a healthy and sound financial situation,and have the ability to effectively respond to potential market crises,and avoiding falling into financial distress under the pressure of increasingly fierce market competition,which is very important to listed companies.With the development of computer technology and artificial intelligence,based on the original statistical type of corporate financial distress model,artificial intelligence is used to study corporate financial distress,and the establishment of a self-learning artificial intelligence enterprise financial distress early warning model has become the latest research trend.According to the company's operating conditions,building a practical and effective system for predicting and analyzing the financial distress of listed companies can help all stakeholders of listed companies to obtain effective corporate financial early warning and corporate prospect analysis.It can promote the healthy and orderly development of our country's market economy.This paper selects the financial data of all A-share listed companies from 2015 to2017 from the Oriental Wealth Choice financial terminal as the research object.Based on data mining and machine learning algorithms,establishes an Adaboost-DEGWO-SVM combined financial distress prediction model.The main research contents of this paper are as follows:(1)Sample data set preprocessing.Firstly,explain the source of the sample data and the principle of selection,then deal with the missing values and outliers of the sample data set.Finally,in order to have a certain numerical comparison of financial data of different magnitudes,the original data was normalized to construct a 1:1distressed company(ST)and a normal company to form a modeling data set.(2)Extract modeling properties.Firstly,the pearson correlation coefficient algorithm is used to eliminate the collinear financial indicators.Under the condition of ensuring the integrity of the original data set,the principal component analysis algorithm(PCA)is used to further condense the indicator variables and find out 13 important financial indicators as the modeling attributes.(3)Construct Adaboost-DEGWO-SVM combined prediction model.Firstly,optimize the parameter optimization process of the SVM algorithm through the gray wolf optimization algorithm(GWO).Then,the differential evolution algorithm(DE)is used to improve the global search ability of gray wolf optimization algorithm(GWO),so as to solve the problem of the gray wolf algorithm that is easy to fall into local optimization.Finally,the Adaboost algorithm is used to improve the classification ability of the DEGWO-SVM model,and realize the construction of the Adaboost-DEGWO-SVM combined model.The experimental results show that the Adaboost-DEGWO-SVM combined forecasting model proposed in this paper has obvious advantages in predicting corporate financial distress.Compared with the DEGWO-SVM model,the classification accuracy is increased by 4.34%,and the type I and type II errors are reduced 0.0435 respectively,compared with the single SVM model,the classification accuracy is increased by 13.04%,the type I error and the type II error are reduced0.1304 and 0.1305 respectively.It is a promising method for predicting corporate financial distress.