Application Of Random Forest In The Credit Risk Evaluation Of Listed Companies In Manufacturing Industry

Scholars at home and abroad have regarded the credit risk and its measurement model as the research frontiers of economic,financial and other fields.With economy of our country stepping into the "new normal" period,manufacturing industry is facing the macro policy adjustment,the industrial transformation and upgrading,and a series of challenges.Manufacturing industry is not only the pillar industry of the real economy,but also occupies half part in the capital market.Financial market in our country is improving day by day,the enterprises which incur consecutive losses will be abandoned by the market and investors.Therefore,the establishment of the credit risk evaluation model not only can enrich the research in the field,but also can give more reliable judgment basis to inform users.Based on the findings about random forests and credit risk problem studied by Chinese and foreign scholars,we choose 160 listed enterprises' interim reports which include 26 financial indicators of five dimensions.Research ideas includes the outlier processing,the construction of index system and the establishment of the random forest model and related parameters optimization,we divide the sample into training set and test set,and use the training set for model training and the test set for the model predict,and conclude that random forests have better classification accuracy.In order to better reflect the advantage of random forest in deal with the problem of target industries' credit risk,we use Random Forest?CART model and the SVM model to predict the same datum.Through comparing and analyzing the predictions of three models,we find that RF is significantly better than others.In general,it has significant influence to the theory research and real economics by using RF model to quantify and warn risk.