基于随机森林的企业信用风险评估模型研究

Research on Enterprise Credit Risk Assessment Model based on Stochastic Forest

随着信贷规模的扩张,企业信用风险评估成为银行重点关注的问题。本文首先梳理企业信用风险评估文献,然后引入非财务指标,对现有指标体系进行完善;在此基础上,利用随机森林方法建立企业信用风险评估模型;并从指标类型和评估方法两个角度对所建模型进行评价。结果表明:盈利能力、偿债能力以及管理层激励对企业信用风险影响较大;改进后的指标体系能显著提高模型预测准确率;随机森林的预测性能优于CART决策树。

With the expansion of credit scale, enterprise credit risk assessment has become a major concern of banks. Based on the literature review of enterprise credit risk assessment, this paper introduces non-financial indicators to improve the existing index system; on this basis, taking the listed companies as the research object, uses random forest method to establish enterprise credit risk assessment model; evaluates the model from two angles of index type and evaluation method. The result shows that the profitability, solvency and management incentive have great influence on the credit risk of the enterprise; the improved index system can significantly improve the forecast accuracy of the model; the prediction performance of the random forest is superior to the CART decision tree.