基于集成极限学习机的电子商务风险预警研究

Research of E-commerce Risk Warning Based on Ensemble Extreme Learning Machine

有效的信用风险预警可以降低电商商务活动中的风险,促进电子商务的发展。以极限学习机为基分类器的集成模型适用于电子商务企业信用数据样本少,维度高的特点;通过对少数类样本过采样缓解类别不平衡问题,进一步提高模型预测准确率。实证分析表明,基于类别平衡校正的集成极限学习机能够对企业风险预警等级作出有效的预测,且结果优于现有的基于传统机器学习算法及对应的集成模型,对提升电子商务风险预警效果有积极作用。

Effective credit risk warning can reduce the risk in the process of e-commerce business and promote the development of e-commerce. The ensemble model with an extreme learning machine as the base classifier, suitable for e-commerce enterprises with few samples and high dimensionality of credit data, further improves the prediction accuracy of the model by alleviating the class imbalance problem by oversampling a few class samples. The empirical analysis shows that the ensemble extreme learning machine based on class balance correction can make effective predictions of enterprise risk warning levels, with the results better than the existing traditional machine learning algorithms and corresponding ensemble models, and improve the effect of e-commerce risk warning.