Prediction And Optimization Of Default Risk Of Chinese Bonds Based On Light Gbm

In recent years,the development of China’s credit bond market is very rapid,until now,the problem of credit bond default is more and more frequent,some leading enterprises and state-owned enterprises have defaulted,which has not only aroused widespread concern of the society,but also triggered people’s concern about the healthy development of the credit bond market in the future.How to make use of the current level of information disclosure to carry out a more effective risk assessment of individual credit debt is of great significance for controlling credit risk,enhancing investor confidence,standardizing enterprise operation and even promoting the healthy development of China’s bond market.In recent years,the convenience brought by the rapid development in the field of artificial intelligence makes the word machine learning appear more and more frequently in the study of various classical problems in various fields.As a kind of machine learning,random forest algorithm has been widely used in the study of credit default in recent years.Light GBM algorithm,as a new machine learning algorithm,it has some differences with random forest algorithm in principle,but has some similarities in application and principle.This thesis aims to test the validity of the model by using the machine learning method,according to the historical data of the Chinese bond market and the research results of the existing literature on the influence factors of the credit debt default.By studying the research literature on the default of credit debt,this thesis selects the current research on the problem of bond default in the latest machine learning method.Firstly,the method of machine learning is compared,and the Light GBM algorithm is used as the research tool for the prediction effect of the boost algorithm used by many researchers.Then the index system is constructed from three angles: macro-factor,financial factor and non-financial factor.Then,all the relevant characteristics are summarized of the default credit debt that are available except for the 17 Chenglong 03,by using the light GBM algorithm to build the preliminary model of the default prediction of bond defaults through machine learning.Then,according to the characteristics of Light GBM,the index system is optimized,and the final prediction model is constructed based on the optimized index system.Finally,the validity of the model is verified by using the validity of the model 17 Chenglong 03,which proves the validity of the model,and summarizes the conclusions and implications of the thesis,and puts forward some suggestions for the risk control of the default risk in our country’s bond.