基于支持向量机集成学习方法的高新技术上市公司绩效预测研究

A study of performance prediction of hi-tech enterprises: a study based on SVM ensemble learning

本文以创业板高新技术企业为对象,采集样本企业2011-—2014年的年报数据,通过构建支持向量机预测模型和支持向量机集成学习方法预测模型,预测样本企业的绩效。研究结果显示:两种预测模型均能有效预测样本企业的绩效,支持向量机集成学习方法预测模型的准确度更高,样本数据越接近预测年度,预测的准确度越高。

In this paper,we predict the performance of the sample hi-tech enterprises from GEM on the basis of their2011-2014 financial data. To complete the study,we build support vector machine(SVM)prediction model andSVM ensemble lea,rning prediction model. The results show that both models can predict the performance of thesample enterprisesSVM ensemble learning prediction model has higher prediction accuracy,and that the closerthe data are to the forecast year,the higher the prediction accuracy is.