摘要
研究了两类商品期货价格的预测问题。商品期货价格预测是一个时变、动态和非线性问题,价格的变化受到市场及外部金融环境复杂变化条件的影响,具有很强的随机性和非线性特性,成为对商品期货价格预测问题的难点所在。考虑到传统的线性关系模型在处理商品期货价格非线性特征上的不足,以及单一预测模型存在对已知数据空间不能充分学习的问题,提出了一种改进的梯度提升决策树方法实现对商品期货价格的预测分析。同时根据梯度提升决策树方法的预测性能受学习率等参数的影响,提出使用遗传算法对模型参数进行寻优的改进预测模型。仿真结果表明,改进的梯度提升决策树模型的预测性能明显优于对比预测模型,为期货产品的价格预测问题提供了新的预测工具,有效的提高了价格预测的准确度。
Two types of commodity futures prices are studied. Commodity futures price forecasting is a time-varying, dynamic and nonlinear problem, the change of the price is affected by complicated changing conditions of the market and the external financial environment, which has strong randomness and nonlinear characteristics. This is also the difficulty of commodity futures price forecast. Considering the linear relationship between the traditional model in dealing with the nonlinear characteristics of the commodity futures prices, a single prediction model can not fully learn the knowledge containing in dataset. An improved gradient decision tree method was proposed to predict the commodity futures price. At the same time, according to the influence of the learning rate and other parameters on the prediction performance of the gradient promotion decision tree method, genetic algorithm was used to optimize the model parameters. The simulation results show that the performance of the improved gradient decision tree model is better than other prediction models, which provides a new tool for forecasting the price of futures products and improves the prediction accuracy effectively.
作者
罗光华
陈江
柯翔敏
LUO Guang-hua;CHEN Jiang;KE Xiang-min(Information Department,Huaqiao University,Xiamen Fujian 361021,China)
出处
《计算机仿真》
北大核心
2018年第9期421-426,共6页
Computer Simulation
基金
福建省财政科学基金(JA15024)
关键词
梯度提升决策树
遗传算法
集成模型
价格预测
Gradient boosting decision tree
Genetic algorithm
Ensemble model
Price forecasting
