Prediction of Enterprise Credit Bond Default based on Random Forest and Neural Network

Abstract

With the outbreak of many major default events in China's bond market, the problem of corporate bond default prediction has increasingly become the focus of academic and practical circles, and machine learning and deep learning algorithms have been widely used. By constructing a random forest-BP neural network model and a random forest model based on convolutional neural network, we make an empirical analysis of the financial data of listed companies whose bonds default or are due to be paid from 2014 to 2019, and compare the prediction results of the algorithms to test the effectiveness of the models. The empirical results show that the random forest model based on convolutional neural network is the best in the prediction of corporate credit default, and the combined model has better performance than the single model.