Isomap-ABC-RVM预测模型的构建及应用

发布时间：2018-03-05 15:35

本文选题：Isomap-ABC-RVM模型　切入点：预测　出处：《统计与决策》2015年16期 　论文类型：期刊论文

【摘要】：使用机器学习算法建立模型对实际经济问题进行预测分析,是近年来一个研究热点。传统的预测模型使用支持向量机(SVM)并结合一些传统降维算法和参数寻优算法来构建,先用降维算法对样本进行降维,避免样本特征指标相关性高,再通过支持向量机对样本进行机器学习来进行回归预测,而向量机中核参数的寻优则用参数寻优算法来完成。文章引入更为前沿的算法,使用等距特征映射(Isomap)进行降维,人工蜂群算法(ABC)进行参数寻优,相关向量机(RVM)进行回归预测,构建Isomap-ABC-RVM模型并对沪深300指数进行预测分析。实证分析表明Isomap-ABC-RVM具有良好的预测精度,为股指预测提供了新方法。
[Abstract]:It is a hot research topic in recent years to use machine learning algorithm to build models to forecast and analyze actual economic problems. The traditional prediction models are constructed by using support vector machine (SVM) and some traditional dimensionality reduction algorithms and parameter optimization algorithms. First, the dimension reduction algorithm is used to reduce the dimension of the sample to avoid the high correlation of the characteristic index of the sample, and then the regression prediction is carried out through the machine learning of the sample by support vector machine (SVM). The kernel parameter optimization in vector machine is accomplished by parameter optimization algorithm. This paper introduces a more advanced algorithm, using isometric feature mapping (Isomapa) to reduce dimension, artificial bee colony algorithm (ABC) for parameter optimization, and correlation vector machine (RVM) for regression prediction. The Isomap-ABC-RVM model is constructed and the Shanghai and Shenzhen 300 index is forecasted and analyzed. The empirical analysis shows that Isomap-ABC-RVM has good prediction accuracy and provides a new method for stock index prediction.
【作者单位】：武汉大学经济与管理学院;
【分类号】：F724.5;F224

【参考文献】