Article
Operations Research & Management Science
Shaoze Cui, Dujuan Wang, Yunqiang Yin, Xin Fan, Lalitha Dhamotharan, Ajay Kumar
Summary: This study proposes a two-stage heterogeneous ensemble method for predicting carbon trading prices. By extracting multiple feature sets and using various algorithms to construct the model, the research shows that this method outperforms other methods in predicting carbon trading prices.
ANNALS OF OPERATIONS RESEARCH
(2022)
Article
Green & Sustainable Science & Technology
Yuanzhuo Du, Kun Zhang, Qianzhi Shao, Zhe Chen
Summary: This study proposes a novel hybrid prediction method for improving the prediction accuracy of wind power generation. The method includes data correlation analyses, power decomposition and reconstruction, and a new prediction model.
Article
Business, Finance
Kunliang Xu, Weiqing Wang
Summary: A reliable crude oil price forecast is crucial for market pricing. This study incorporates a rolling window into two prevalent EEMD-based modeling paradigms to improve accuracy. The results show that EEMD plays a weak role in improving crude oil price forecasts when only the in-sample set is preprocessed, but the rolling EEMD-denoising model has an advantage for long-term forecasting.
INTERNATIONAL REVIEW OF FINANCIAL ANALYSIS
(2023)
Article
Multidisciplinary Sciences
Ibrahim Alshourbaji, Na Helian, Yi Sun, Abdelazim G. Hussien, Laith Abualigah, Bushra Elnaim
Summary: This paper introduces an enhanced gradient boosting model (EGBM) for churn prediction in telecommunications, which improves the initial classification performance of traditional GBM by using support vector machine and exponential loss function. A modified particle swarm optimization method is also developed to effectively tune the hyper-parameters of CP-EGBM. Evaluation results on seven open-source CP datasets demonstrate that CP-EGBM outperforms other models and shows promising improvements in churn prediction.
SCIENTIFIC REPORTS
(2023)
Article
Public, Environmental & Occupational Health
Dost Muhammad Khan, Muhammad Ali, Nadeem Iqbal, Umair Khalil, Hassan M. M. Aljohani, Amirah Saeed Alharthi, Ahmed Z. Z. Afify
Summary: In this article, a new hybrid time series model called EEMD-ETS is proposed to predict COVID-19 daily confirmed cases and deaths. The model decomposes the complex data into different components and checks their stationarity, resulting in accurate predictions. The model outperforms other time series and machine learning models, making it a recommended choice for COVID-19 prediction.
FRONTIERS IN PUBLIC HEALTH
(2022)
Article
Thermodynamics
Guohui Li, Zhiyuan Ning, Hong Yang, Lipeng Gao
Summary: A new carbon price prediction model is proposed in this paper, which processes and analyzes the data using multiple methods to accurately predict the carbon price, verifies the effectiveness of the model, and can be used to predict the supply and demand of the carbon market and evaluate the effectiveness of carbon trading policies.
Article
Computer Science, Artificial Intelligence
Yuqi Guo, Jianfeng Guo, Bingzhen Sun, Juncheng Bai, Youwei Chen
Summary: This study proposes a novel stock price forecasting model based on system clustering method and particle swarm optimization. By employing ensemble empirical mode decomposition and sample entropy method, stock prices are decomposed into sub-series with different features, and predictions are made using optimized neural networks or regression models, enhancing forecasting accuracy.
APPLIED SOFT COMPUTING
(2022)
Article
Mathematical & Computational Biology
Ravichandra Madanu, Farhan Rahman, Maysam F. Abbod, Shou-Zen Fan, Jiann-Shing Shieh
Summary: A survey found that 47% of surgical complication mortalities are due to anesthetic overdose, highlighting the necessity for regulating anesthesia levels. Deep learning methods have been utilized for predicting patient anesthesia depth using EEG signals, with CNN being a popular algorithm. Various decomposition methods were used in this study to extract features from EEG signals, demonstrating potential for further research in visual mapping of DOA using EEG signals and DL methods.
MATHEMATICAL BIOSCIENCES AND ENGINEERING
(2021)
Article
Environmental Sciences
Wei Sun, Chang Xu
Summary: A novel hybrid forecasting model, incorporating EEMD, LDWPSO, and wLSSVM algorithms, was proposed for predicting carbon prices. The model demonstrated the best forecasting performance among all model combinations in the evaluation across three regions, significantly enhancing the accuracy of carbon price prediction.
SCIENCE OF THE TOTAL ENVIRONMENT
(2021)
Article
Computer Science, Hardware & Architecture
Cry Kuranga, Njodzi Ranganai, Tendai S. Muwani
Summary: Real-world nonstationary data are usually characterized by high nonlinearity and complex patterns, making prediction a challenging task. In this work, a dynamic particle swarm optimization-based empirical mode decomposition ensemble is proposed to improve prediction accuracy. Experimental results on electric time series datasets show that the proposed technique outperforms several state-of-the-art techniques.
JOURNAL OF SUPERCOMPUTING
(2022)
Article
Business
Kunliang Xu, Hongli Niu
Summary: This study proposes a new ensemble model based on sliding decomposition for crude oil futures price prediction. The empirical findings indicate that this model does not improve prediction accuracy compared to other models in out-of-sample data.
TECHNOLOGICAL FORECASTING AND SOCIAL CHANGE
(2022)
Article
Engineering, Multidisciplinary
Jinde Zheng, Miaoxian Su, Wanming Ying, Jinyu Tong, Ziwei Pan
Summary: The study introduces the improved Uniform Phase Empirical Mode Decomposition (IUPEMD) method, which enhances the accuracy and performance of signal decomposition by adaptively selecting the amplitude of the sinusoidal wave and choosing the optimal result based on orthogonality index.
Article
Energy & Fuels
Daniel Marc Banks, Johannes Cornelius Bekker, Hendrik Johannes Vermeulen
Summary: Parameter estimation is an important aspect in modeling electromagnetic systems, with various strategies explored in literature. Most methodologies use time-domain or frequency-domain responses from the device under test. However, limited research has been done on using modal decomposition strategies for parameter estimation using time-domain waveforms. This paper explores the use of Empirical Mode Decomposition for estimating the parameters of a three-section lumped parameter transformer model. A novel approach is proposed to define the optimization cost function based on the intrinsic modes of simulated time-domain waveforms, and results are compared with classical time-domain and frequency-domain approaches.
Article
Biochemistry & Molecular Biology
Ying Zhou, Erteng Jia, Huajuan Shi, Zhiyu Liu, Yuqi Sheng, Min Pan, Jing Tu, Qinyu Ge, Zuhong Lu
Summary: A method called Multi-LSTM based on signal decomposition technique and deep learning is proposed for predicting gene expression levels, which efficiently reduces data nonlinearity and improves prediction robustness according to experimental results.
INTERNATIONAL JOURNAL OF MOLECULAR SCIENCES
(2022)
Article
Multidisciplinary Sciences
Libiao Chen, Qiang Ren, Juncheng Zeng, Fumin Zou, Sheng Luo, Junshan Tian, Yue Xing
Summary: The implementation of toll-free during holidays leads to a significant increase in traffic congestion on expressways. Real-time and accurate holiday traffic flow forecasts are crucial for guiding diversion and reducing congestion. However, most current prediction methods focus on regular working days or weekends, and there are limited studies on predicting traffic flow during festivals and holidays. Therefore, a data-driven expressway traffic flow prediction model based on holidays is proposed, which effectively captures the spatial-temporal correlation and heterogeneity of traffic flow components using the Spatial-Temporal Synchronous Graph Convolutional Networks (STSGCN) model. The model achieves excellent results in predicting fluctuating holiday traffic flow, offering valuable insights for future travel choices and road network operation.