Research On Green Credit Default Risk Prediction Of Listed Companies Based On Ensemble Algorithm

Sustainable development is an important principle of current economic and social development,but the rapid economic development is accompanied by energy consumption and environmental pollution.In order to achieve sustainable development,China has made a lot of efforts in the field of ecological civilization construction,among which green finance is an important manifestation of promoting sustainable development in the financial field.As an important part of green finance,green credit plays a very important role in promoting sustainable economic development.Compared with developed countries,the development of green credit in China started late,and the methods and research of green credit risk identification and measurement are not mature enough,which restricts the development of green credit business in China.Therefore,based on the analysis of the development status and risks of green credit in China,this paper uses the integrated algorithm to build a green credit default risk prediction model of listed companies,providing effective basis for the improvement of green credit risk assessment index system and green credit risk prevention and control.This paper first introduces the relevant policies of green credit,the status quo and development of green credit.Secondly,the connotation characteristics and risk sources of green credit risk are sorted out.Then,a model incorporating corporate social responsibility and corporate environmental responsibility indicators is constructed,and an empirical study on green credit default risk of listed companies is carried out using the integrated algorithm.According to the recent years the credit rating of listed companies can be divided into high and low credit risk two kinds of enterprise credit risk.The financial indicators,corporate characteristics,social responsibility and environmental responsibility indicators of listed companies in 2019 are selected as the initial indicators.Data is preprocessed using Python,through the random forest algorithm,XGBoost algorithm and Cat Boost algorithm after parameter tuning,the green credit default risk of listed companies can be classified and predicted.The prediction model was improved by using the mean importance of indicators generated in model fusion and model construction of three integration algorithms to eliminate the indicators with lower importance.Finally,the accuracy rate and recall rate of the optimal fusion model reached 96.61% and 96.94%.Based on the established model,the green credit default risk of listed companies in the environmental protection industry and the environmental protection industry is compared.Finally,based on the above research,relevant suggestions are put forward for the government,banks and enterprises to prevent green credit risks,so as to promote the sustainable development of green credit and green finance in China.