Research On Financial Crisis Early Warning Of Advanced Manufacturing Enterprises Based On Machine Learning

With the high attention and support of the state,in China’s financial market,advanced manufacturing enterprises have grown and expanded in number and scale,and they will also bear dangers and challenges while facing opportunities,and the increasingly active market competition may lead to financial crises for enterprises.Therefore,this thesis constructs a scientific and reasonable early warning index system for advanced manufacturing enterprises,and constructs an early warning model with better performance.Firstly,this thesis selects 13 sub-industries as advanced manufacturing industry according to various documents,selects sample enterprises on this basis,obtains the financial indicators of sample enterprises through the database,tests and screens them,and uses mutual information and principal component analysis to reduce dimensionality respectively,so that there is no redundancy between the screened indicators and there are obvious differences between the two types of enterprises,as a financial index system.Secondly,in order to make up for the shortcomings of the lag of financial indicators,text indicators are specially introduced to supplement them,by combining the characteristics of text information,the crawler technology is used to select the review data of Oriental Wealth Stock Bar,data cleaning,Chinese word segmentation and stop word processing,and the dictionary method is used to analyze the sentiment of the processed data,and in order to improve the accuracy;the financial emotional dictionary is constructed through the SOPMI algorithm to supplement the basic dictionary.Then,according to the sentiment value calculation rules set in this thesis,the sentiment score of each enterprise is obtained as a text indicator.Finally,based on the principle of model,K-fold cross-verification is introduced when the longhorn ox must search algorithm optimizes SVM,and then the BAS-SVM model is constructed,and then the financial indicators after mutual information and PCA dimensionality reduction and the combined index system of text indicators are used as input values to predict the financial crisis of advanced manufacturing enterprises,and compare and analyze with other models to show the perfection of the early warning index system after the introduction of text indicators and the accuracy of optimization model prediction,Select an early warning model with good prediction effect,and make suggestions to relevant personnel from the perspective of economics and management.The results show that the BAS-SVM model proposed in this thesis has good predictive performance,which enriches the application of SVM model in the prediction of financial crisis of advanced manufacturing enterprises,in addition,the data quality after PCA dimensionality reduction processing is higher than that of mutual information screening,and the introduction of text indicators greatly improves the prediction performance of the model.