基于机器学习的多模型黄金股票价格预测.pdfVIP

长，信息通信

DOI:10.20153/j.issn.2096-9759.2025.02.050

基于机器学习的多模型黄金股票价格预测

倪文周，田嫚，焦一清，茹慧英

河北建筑工程学院理学院，河北张家口075000

DynamicTrendAnalysisofGoldPricesBasedonMachineLearningModels

NIWenzhou,TIANman,JIAOYiqing,RUhuiying

SchoolofScience,HebeiUniversityofArchitecture,Zhangjiakou075000,Hebei,China

摘要：黄金价格股票作为经济分析领域Abstract:Asapivotalindicatorineconomicanalysis,goldpricesplayabarometerroleinnation-

的重要指标，对国民经济的盈亏起着风alcconomicfluctuations:Predictivemodelsforgoldpricescnableproactiveriskmitigationin

向标的作用。黄金价格预测模型有助于marketeconomies.Thisstudycomparativelyanalyzesthreepredictivemodels:theARIMAtime

国民在市场经济下能够占据提前避险优seriesmodel,theXGBoostensemblelearningalgorithm,andtheLSTMdeeplearningmodel,e-

势。文章分别选用时间序列ARIMA模valuatedthroughRMSE,MAE,andMAPEmetrics.Resultsdemonstratethesuperiorperform-

型、集成学习算法中的XGBoost模型以anceoftheLSTMmodel(2.194,1.789,0.73%)comparedtoARIMA(2.62,2.33,0.83%)

及深度学习中的LSTM模型对黄金股票andXGBoost(6.161,4.943,0.83%).Toenhancepredictionaccuracy,weimplementeddeper

价格进行预测。通过对比三者的RMSE、networkarchitecturesandbidirectionalLSTM,achievingoptimizedmetricsof1.337,1.057,

MAE、MAPE三项模型指标，发现应用中and0.42%respectively.Thisadvancementsignificantlyimprovesdynamicgoldpricepattern

的LSTM时间序列模型更优，其中，ARI-recognition,offeringenhancedpredictivecapabilitiesforeconomicdecision-making.Themeth-

MA模型三项指标为2.62、2.33、0.83%；

odologicalprogressionfromtraditionalstatisticalmodelstosophisticatedneuralarchitecturesun-

XGBoost模型为6.161、4.943、0.83%;

derscorestheeffectivenessofdeeplearningapproachesinfinancialtimeseriesforecasting.

LSTM模型为2.194、1.789、0.73%。为