基于主成分的RBF核SVM在财务预测领域的应用

Application of Principal Component Based RBF Kernel SVM in Financial Forecasting

支持向量机(SVM)是实现结构风险最小化归纳原理的一种机器学习理论,在有限的学习模式下具有良好的泛化能力。为了评估支持向量机的预测性能,本文通过对684家企业进行财务分析,进而预测企业在未来两年是否会被ST。建立基于主成分的RBF(核函数)核SVM模型,将支持向量机与传统学习算法进行比较,结果表明支持向量机有效地提高了预测的精度,具有良好的泛化和预测能力。

Support vector machine(SVM) is a machine learning theory to minimize structure risk,which performs good generalization ability in limited learning mode.To evaluate the predictive performance of support vector machine,this paper predicts whether the enterprise will be subject to ST in the next two years through financial analysis of 684 enterprises. The RBF kernel SVM model based on principal components was established and we compare it with traditional learning algorithms.The results show that SVM can improve the prediction accuracy effectively and has good generalization and prediction ability.