A novel deep learning framework: Prediction and analysis of financial time series using CEEMD and LSTM

Zhang, Yongan; Yan, Binbin; Memon, Aasma

doi:10.1016/j.eswa.2020.113609

Cited by 140 publications

(46 citation statements)

References 55 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Integrating data processing techniques into data-driven models can effectively improve the prediction accuracy, which is the joint conclusion of many previous studies, such as [34]- [39]. A common workflow adopted in these studies is the decomposition-ensemble framework, which decomposes the original time series and reconstructs the final time series prediction.…”

Section: Adaptive Decomposition-ensemble Frameworkmentioning

confidence: 96%

“…Data processing techniques e.g., empirical wavelet transform (EWT) [29], empirical mode decomposition (EMD) [30], wavelet packet decomposition (WPD) [31], singular spectral analysis (SSA) [32], and variational mode decomposition (VMD) [33], can not only extract multiscale features from time series but can also make the prediction easier [34], [35]. The coupling of data-driven models and data processing techniques has occurred in diverse studies, including runoff forecasting [36], wind speed forecasting [37], solar photovoltaic power forecasting [38], and stock index forecasting [39]. In these studies, data processing techniques decompose the whole signal into multiple subsignals, and data-driven models make predictions for each of the subsignals.…”

Section: Introductionmentioning

confidence: 99%

“…In a real engineering case, this adaptive framework can avoid misuse of future inflow information, which allows the developed model to adapt to the real inflow observation and prediction processes [40]. Here, VMD and LSTM are incorporated into the above framework for signal decomposition and prediction, respectively, because of their robust performance [9], [29], [39]. Finally, the practicability of the developed model in a real engineering problem is tested on inflow forecasting of the Three Gorges Reservoir (TGR).…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Adaptive Reservoir Inflow Forecasting Using Variational Mode Decomposition and Long Short-Term Memory

Yang

2021

IEEE Access

View full text Add to dashboard Cite

Developing an accurate and timely inflow forecast model continues to be an integral part of modern reservoir operation and has been greatly promoted by the combination of data preprocessing and data-driven techniques. This paper integrates variational mode decomposition (VMD) and long short-term memory (LSTM) into a hybrid model named A-VMD-S-LSTM. After dividing data for calibration and validation periods, A-VMD-S-LSTM decomposes the calibration inflow time series into subseries and cast them into a single LSTM forecast model to output final inflow predictands directly. When new inflow data are appended to the calibration time series, the VMD and LSTM models will be updated to adapt to all available inflow information to avoid transmitting future inflow information. The parameter settings of VMD and LSTM and the meaningful lags corresponding to each subseries are well identified by grid search and partial autocorrelation function (PACF), respectively. Finally, the proposed model is applied to the 1-, 3-5-, and 7-day ahead inflow forecasts of the Three Gorges Reservoir (TGR) along with five competitive models. Results show that A-VMD-S-LSTM achieves the best forecast accuracy and fourth computational efficiency among the six candidates, demonstrating its best comprehensive forecast performance. Furthermore, using two or three LSTM layers in A-VMD-S-LSTM is redundant since the change in forecast accuracy is negligible and the increase of run time is remarkable. Therefore, a practical reservoir inflow forecast model with high forecast accuracy and relatively low computational cost is provided in this paper. INDEX TERMSReservoir inflow, adaptive forecasting, variational mode decomposition, long short-term memory, parameter calibration.

show abstract

Section: Adaptive Decomposition-ensemble Frameworkmentioning

confidence: 96%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Adaptive Reservoir Inflow Forecasting Using Variational Mode Decomposition and Long Short-Term Memory

Yang

2021

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Accordingly, machine learning models, such as the support vector machine and artificial neural network (ANN) models, have been applied to forecast the value or direction of the stock market index to overcome the shortcomings of linear models [10]- [12]. Recently, deep learning models, including long short-term memory (LSTM) [13]- [18], and their variants [19]- [25] have been popularly proposed for stock prediction.…”

Section: Introductionmentioning

confidence: 99%

Forecasting Stock Market Indices Using Padding-Based Fourier Transform Denoising and Time Series Deep Learning Models

2021

View full text Add to dashboard Cite

Approaches for predicting financial markets, including conventional statistical methods and recent deep learning methods, have been investigated in many studies. However, financial time series data (e.g., daily stock market index) contain noises that prevent stable predictive model learning. Using these noised data in predictions results in performance deterioration and time lag. This study proposes paddingbased Fourier transform denoising (P-FTD) that eliminates the noise waveform in the frequency domain of financial time series data and solves the problem of data divergence at both ends when restoring to the original time series. Experiments were conducted to predict the closing prices of S&P500, SSE, and KOSPI by applying data, from which noise was removed by P-FTD, to different deep learning models based on time series. Results show that the combination of the deep learning models and the proposed denoising technique not only outperforms the basic models in terms of predictive performance but also mitigates the time lag problem.

show abstract

“…Compared with traditional time-frequency domain and wavelet packet analyses, the adaptive decomposition method is more intuitive and adaptive given that no basis function needs to be preset. Accordingly, this method has attracted wide usage used in the field of fault diagnosis [8][9][10]. Empirical mode decomposition (EMD) is a recursive mode decomposition method that uses the extreme points of the original signal to perform multiple envelope calculations.…”

Section: Introductionmentioning

confidence: 99%

Combine Assembly Quality Detection Based on Multi-Entropy Data Fusion and Optimized LSSVM

Zhao

Zhang

et al. 2021

IEEE Access

View full text Add to dashboard Cite

To solve the problems related to the complex structures, multiple parts, and imperceptible assembly quality of combines, this paper compares the performance of the empirical mode decomposition (EMD), ensemble empirical mode decomposition (EEMD), complete ensemble empirical mode decomposition (CEEMD), complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) and information entropy features in the detection data of combine assembly quality, and proposes a vibration detection method of combine assembly quality detection based on multi-entropy feature fusion and optimized least squares support vector machine. Firstly, the vibration signals of the combine are decomposed by various adaptive algorithms, and intrinsic modal function (IMF) components are obtained. Three entropy features are extracted from the components of the modes. The features are visualized by t-distributed stochastic neighbor embedding (t-SNE), and the performance of these entropy features in the combine assembly quality detection is analyzed. Secondly, a feature extraction method based on information entropy fusion is proposed. The optimized kernel principal component analysis (KPCA) is used to fuse and reduce the dimension of the entropy features, and form the fusion features. Finally, the extracted features are imported into optimization least squares support vector machine (LSSVM) model for training to judge the working state and assembly quality problem type of combine. The results show that the accuracy of using the unfused entropy features is 82.5%, the accuracy of the fused features is 87.9%. After WOA optimization, the accuracy of the final classification reaches 92.1%, which is better than 89.6% by GA and 90.5% by PSO. It shows that the method proposed in this paper can accurately identify combine problems with different conditions and can be applied to combine assembly quality detection.

show abstract

A novel deep learning framework: Prediction and analysis of financial time series using CEEMD and LSTM

Cited by 140 publications

References 55 publications

Adaptive Reservoir Inflow Forecasting Using Variational Mode Decomposition and Long Short-Term Memory

Adaptive Reservoir Inflow Forecasting Using Variational Mode Decomposition and Long Short-Term Memory

Forecasting Stock Market Indices Using Padding-Based Fourier Transform Denoising and Time Series Deep Learning Models

Combine Assembly Quality Detection Based on Multi-Entropy Data Fusion and Optimized LSSVM

Contact Info

Product

Resources

About