基于支持向量机的网络贷款平台等级评价

采用有监督学习方法建立支持向量机模型,对网络贷款平台的等级评价方法进行了研究。由于学习样本少,为了提高模型的泛化能力,采用K-fold交叉验证方法进行模型学习。采用不同的核函数分别建立6种支持向量机模型,选择经过5轮交叉验证后正确率最高的模型作为最终的输出模型。验证结果表明,虽然学习样本数量不多,所建立的支持向量机模型在验证集中预测准确度达80%。

Level Evaluation of Peer-to-peer Lending Platform Based on Support Vector Machine

A supevived support vector machine was applied to evaluate risk level of network loan platforms. Generalization ability was improved By using K-fold cross validation method in the case of small learning sample size. We developed six support vector machine models, based on different kernel functions. After 5-fold cross validation, the linear model with the higest identification accuracy was selected as the output model. The model was used to verify samples. The precision of prediction was up to 80%.