基于Lasso-PSO-BP方法的中国黄金期货价格短期预测  

Short-term Forecast of Chinese Gold Futures Price Based on Lasso-PSO-BP Method

作　　者：尹晨曦 YIN Chenxi(School of Statistics,Lanzhou University of Finance and Economics,Lanzhou 730020,China)

机构地区：[1]兰州财经大学统计学院,甘肃兰州730020

出　　处：《洛阳理工学院学报（自然科学版）》2022年第4期81-87,共7页Journal of Luoyang Institute of Science and Technology：Natural Science Edition

摘　　要：针对中国黄金期货价格非线性、非平稳性、高波动性等特征,根据黄金期货相关影响因素,提出一种基于Lasso变量选择的PSO-BP神经网络预测模型。选取影响黄金价格的相关因素,利用Lasso对所选因素进行特征提取,再根据时差相关分析法和相空间重构确定所有变量的滞后期,引入PSO-BP神经网络进行预测。经PSO优化的BP神经网络预测效果要好于单一的LSSVR、BP神经网络与ELM模型,同时也好于PSO-LSSVR和PSO-ELM模型。Aiming at the non-linearity,non-stationary and high volatility of gold futures prices in China,a PSO-BP neural network forecasting model based on Lasso variable selection is proposed according to the relevant influencing factors of gold futures.Firstly,Lasso is used in the process to extract the features of the factors affecting the gold pricesselected.Then,the lag time of all variables is determined by time difference correlation analysis and phase space reconstruction.Finally,PSO-BP neural network is introduced to predict the gold price.The results show that the prediction effect of BP neural network optimized by PSO is better than other networks and models,including LSSVR,BP,ELM,PSO-LSSVR and PSO-ELM.

关 键 词：黄金期货 Lasso 时差相关分析 预测 

分 类 号：O29[理学—应用数学] F832.54[理学—数学]