摘要
首次将遗传规划算法(GP)应用于集成预测问题的研究,使用最小二乘估计算法(LSE)对经典遗传规划算法进行改进,提出了一种基于"遗传规划&最小二乘"算法(GPLSE)的非线性集成预测方法.使用五种不同的模型预测2011年1月-2012年4月青岛港集装箱吞吐量.结果表明,基于GP-LSE的非线性集成预测方法在预测数值准确度和方向准确度两个维度都显著优于其他预测模型.
In this paper, we apply to the genetic programming algorithm （GP） for the ensemble forecast problem, to improve the standard genetic programming algorithm by using least square estimation algorithm （LSE）, and propose a nonlinear ensemble forecasting model based on ＇genetic programming ＆ least square estimation＇ algorithm （GP-LSE）. Five different forecasting models are used to predict throughputs of Port of Qingdao during the period from January 2011 to April 2012. Empirical results show that the nonlinear ensemble forecasting model based on GP-LSE significantly outperforms other competitive forecasting models, in terms of both the level and the direction.
出处
《系统科学与数学》
CSCD
北大核心
2013年第11期1332-1344,共13页
Journal of Systems Science and Mathematical Sciences
基金
国家自然科学基金(71171011
711710112)
中央高校基本科研业务费专项资金(ZZ1133)资助课题
关键词
非线性集成预测
遗传规划
集装箱吞吐量
Nonlinear ensemble forecast
genetic programming
container through-put.
