摘要
工业增加值是衡量一个国家工业发展水平的重要指标。由于其受多种因素影响,对其预测相对困难。本文提出运用时间序列预测方法对其预测,并利用支持向量机和微分进化算法(differential evolution,DE)相结合的方法对中国工业增加值数据进行预测。数据仿真显示该模型比核主成分分析的最小二乘支持向量机(KPCA-LS-SVM)以及岭回归(ridge regression,RR)具有更高的预测精度。
Value-added of industry is an index to measure development of industry in a country. Its forecast is dif- ficult due to too many affecting factors. Methods based on time series are put forward in the paper and applied to forecast, with hybrid method combining support vector machine with differential evolution. The results of data simulation show the method is more accurate than that of KPCA-LS-SVM and ridge regression.
出处
《运筹与管理》
CSCD
2008年第3期88-92,87,共6页
Operations Research and Management Science
基金
国家自然科学重点基金资助项目(70433003)
关键词
工业经济
趋势预测
支持向量机
industrial economics
trend forecasting
support vector machine
