摘要
针对2022年第十四届“华中杯”大学生数学建模挑战赛B题,首先,对44个经济技术指标与“数字经济板块”中的收盘价进行皮尔逊相关性分析,提取出高度相关的指标;然后,使用变分模态分解算法对收盘价历史数据进行分解以降低训练集的非平稳性,再使用长短期记忆神经网络对分解后的各模态分量进行预测并加和重构;其次,对收盘价和与其高度相关的指标进行多元线性拟合得到关系式,并使用粒子群算法优化权重来修正预测模型,采用时间序列交叉验证对模型进行评判,结果表明模型的泛化能力良好;最后,基于收盘价预测结果,结合相关强弱指标、对数移动均线、布林线及夏普比率来进行量化投资.
Aiming at Question B of the 14th"Huazhong Cup"mathematical modeling challenge for college students in 2022,Firstly,Pearson correlation analysis was carried out on a total of 44 economic technical indicators and the closing price in the"digital economy sector"to extract highly relevant indicators;Then,the Variational Modal Decomposition method is used to decompose the historical data of the closing price to reduce the non-stationarity of the training set,and then the long-term and short-term memory neural network is used to predict and reconstruct the decomposed modal components;Secondly,the relationship between the closing price and its highly related indicators is obtained by multivariate linear fitting,and the prediction model is modified by using particle swarm optimization to optimize the weight,and the model is evaluated by time series cross validation.The results show that the generalization ability of the model is good;Finally,based on the closing price prediction results,combined with Relevant strength indicators,Log moving average,Boll line and Sharp ratio to quantify investment.
作者
李佳裕
陈曦
刘闻仲
LI Jiayu;CHEN Xi;LIU Wenzhong(College of Electrical&New Energy,China Three Gorges University,Yichang,Hubei 443002,PR China)
出处
《数学建模及其应用》
2022年第3期72-84,共13页
Mathematical Modeling and Its Applications
关键词
皮尔逊
变分模态分解
长短期记忆神经网络
粒子群优化
交叉验证
量化投资
Pearson
variational mode decomposition
long short-term memory neural network
particle swarm optimization
cross-validation
quantitative investment
