基于改进支持向量机的林业资金投资预测方法  

A novel forestry investment forecast method based on improved support vector machine

作　　者：陶佳[1] 沈红岩[1] 高冠东[2] 

机构地区：[1]河北农业大学信息科学与技术学院,河北保定071000 [2]中央司法警官学院信息管理系,河北保定071000

出　　处：《河北农业大学学报》2012年第4期123-126,共4页Journal of Hebei Agricultural University

摘　　要：针对林业资金投资变化的定量预测,提出一种基于改进支持向量机的预测方法。利用滑动时间窗口方法将历年林业资金投资数据构造成时间序列,将其做为数据样本集并由改进支持向量机加以训练以得到预测模型。通过某省近20年的林业资金投资数据实验验证了预测方法的有效性,实验结果表明:与传统预测方法相比,基于改进支持向量机的预测方法明显提高了投资变化预测精度。Focused on forecast investment forecast, a novel method based on improved support vector machine with particle swarm optimization（SVM-PSO） was proposed, in order to improve forecast accuracy. With sliding time window, history of forestry investment data bad been constructed into a time series. In order to obtain forecast model, the time series was trained as sample set by SVM-PSO. Finally, the experiments, whose data is from the forestry investment of nearly 20 years, show that the SVM-PSO forecast method has better performance than the traditions.

关 键 词：林业资金投资 回归预测 时间序列 支持向量机 粒子群算法 

分 类 号：F326.20[经济管理—产业经济]