基于变分模态分解的极限学习机在金融时间序列预测中的应用

数学理论与应用

基于变分模态分解的极限学习机在金融时间序列预测中的应用

侯木舟¹，骆家辉¹，罗秋懿²

1中南大学数学与统计学院，湖南长沙，410083;
2 中国人民解放军95538部队，四川成都，611400

出版日期:2020-12-30

Application of Extreme Learning Machine Based on Variational Modal Decomposition in Financial Time Series Forecasting

Online:2020-12-30

摘要/Abstract

摘要：

金融时间序列一直以来以其非线性、非平稳、信噪比低等特性成为时间序列预测中的难题。本文提出基于变分模态分解的VMD-ELM模型，利用变分模态分解在复杂的金融时间序列数据分解上的特有优势，将金融时间序列数据分解为若干个子模态，再将分解后的子模态作为极限学习机的输入数据进行训练。基于平均绝对误差（MAE），平均绝对百分比误差（MAPE）和平方根均方根（RMSE），通过比较EMD-ELM模型，前馈神经网络（FFNN）和自回归移动平均（ARMA）在西德克萨斯中质原油（WTI），加拿大/美国汇率（CANUS），美国工业生产（IP）和芝加哥期权交易所纳斯达克100波动率指数（VIX）时间序列数据上的效果，证明了本文提出的方法在多个数据集上均有优秀的表现。

关键词:

金融时间序列, 变分模态分解, 极限学习机

Abstract:

Financial time series has always been a problem in time series forecasting due to its non-linear, non-stationary, and low signal-to-noise ratio. This paper proposes a VMD-ELM model based on variational modal decomposition, and uses the unique advantages of variational modal decomposition in digital signal decomposition to decompose financial time series data into several sub-modes, and then use the decomposed sub-modes as The input data of the extreme learning machine is trained. By comparing the EMD-ELM model, the autoregressive moving average (ARMA) process on the feedforward neural network (FFNN) and the West Texas Intermediate (WTI), the Canadian/US exchange rate (CANUS), and the US industrial production (IP) And the effect on the time series data of the Nasdaq 100 Volatility Index (VIX) of the Chicago Board Options Exchange. Based on average absolute error (MAE), average absolute percentage error (MAPE) and square root mean square (RMSE), the predicted analysis results prove that the method proposed in this paper has excellent performance on multiple data sets.

Key words:

Financial time series, Variational modal decomposition, Extreme learning machine

侯木舟, 骆家辉, 罗秋懿 .

基于变分模态分解的极限学习机在金融时间序列预测中的应用 [J]. 数学理论与应用.