PCR-RBF-SVM预测模型在财政数据中的应用  

Application of Prediction Model Based on PCR-RBF-SVM to Finance Data

作　　者：王喆[1] 王有力 孙雯雯[1] 吕巍[1] 

机构地区：[1]吉林大学计算机科学与技术学院,长春130012 [2]吉林吉信通信咨询设计有限公司,长春130012

出　　处：《吉林大学学报（理学版）》2012年第1期111-113,共3页Journal of Jilin University:Science Edition

基　　金：国家自然科学基金(批准号:60673099;60873146);吉林省科技发展计划重点项目(批准号:20090304)

摘　　要：通过使用支持向量机算法将主成分回归的线性预测结果和径向基神经网络的非线性预测结果相结合,提出一种新的预测模型,该模型提高了预测精度,解决了预测方式单一的问题.将新预测模型应用于财政数据预测结果表明,与传统主成分回归和径向基神经网络方法相比,该模型预测效果更好.On the basis of support vector machine algorithm and the result of principal component regression of the linear prediction and radial basis function neural network of the non-linear prediction,a new forecasting model was proposed by which one can effectively improve the prediction accuracy and solve the problem of single prediction.Application of the new prediction model to the prediction of finance data showed that compared with the traditional principal component regression and radial basis neural network method,the new model has better effect and practical significance in prediction.

关 键 词：主成分回归 径向基神经网络 支持向量机 预测 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]