基于支持向量机的信托公司风险预警指标及模型研究--来自2005年-2013年信托业的经验证据

The Risk Early-Warning Index and Model for Trust Companies Based on Support Vector Machine——Empirical Evidences from the Trust Industry during 2005-2013

信托业作为金融业的重要组成部分,在金融体系服务于实体经济过程中发挥着日益重要的作用。随着我国信托业务规模的不断增大,如何强化我国信托业的风险控制意识,构建我国信托业的风险控制体系,将成为我国金融体系安全建设的重要内容之一。对此,本文分析了支持向量机的建模理论与方法,设计了信托公司风险预警指标,并选取2005—2013年期间信托业的数据样本,通过预警指标赋值方式测算出了样本信托公司的风险测度值;在此基础上,本文进一步运用支持向量机方法,构建了信托公司的风险预警模型,并给出了风险预警模型的分类预警输出结果。研究表明,支持向量机预警模型不仅对于少样本、高维度数据的预警效果较好,而且在分类预警方面具有较高的精确度。本文的研究成果将为我国信托业构建科学、高效的风险预警模型,提升我国信托业的风险运营效能,提供重要的理论指导与决策参考。

Risk management is always an important component of enterprise＇s financial activities. China＇s rapid economic development in recent years surely has huge impacts on the operational and financial environments for enterprises. So how to manage the increasingly sophisticated risks is a key subject for enterprises： This is especially important for China＇ s trust companies which were born with high risks. Support Vector Machine （SVM） is a new measurement of machine-learning on the basis of Statistics Learning Theory （SLT）. It can solve problems like classification and regression efficiently. Because of its high accuracy of classifications dealing with sparse samples data and high dimension data, SVM has become a hotspot of academia. This Paper, attempting to set up a risk early warning index system and model for trust companies based on SVM, intents to provide a more relatively efficient measurement to improve their ability of risk control.