基于机器学习算法的中小企业信用评估研究

Research on credit evaluation of SMEs based on machine learning algorithms

采用大数据方法预测企业违约风险具有重大的现实意义。传统的信用评估模型主要是统计分析模型、判别分析模型等,预测能力有限。因此,文章建立了基于随机森林和支持向量机两种机器学习算法的信用预测模型,并引入ACC、AUC以及FNR评价指标来衡量模型预测的效果。对比实验表明,基于随机森林的信用预测模型较支持向量机模型具有更好的预测效果,证实了模型的优越性。

Using Big Data methods to predict corporate default risk has great practical significance.Traditional credit evaluation models are mainly statistical analysis models,discriminant analysis models,etc.,with limited predictive capabilities.Therefore,this paper establishes a credit prediction model based on two machine learning algorithms:random forest and support vector machine,and introduces ACC,AUC and FNR evaluation indicators to measure the effect of model prediction.Comparative experiments show that the credit prediction model based on random forest has better prediction effect than the support vector machine model,which confirms the superiority of the model.