A denoising carbon price forecasting method based on the integration of kernel independent component analysis and least squares support vector regression

Jianwei, E; Ye, Jimin; He, Lulu; Jin, Hui‐Zi

doi:10.1016/j.neucom.2020.12.086

Cited by 43 publications

(13 citation statements)

References 46 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, due to the nonstationary, nonlinear, and irregular EUA price, it is a particularly difficult issue that requires a more sophisticated approach to analysis. It seems that the solution in this respect could be models using the enoising procedure [5] and deep learning [4]. This will be the subject of further research.…”

Section: Literature Reviewmentioning

confidence: 97%

Ensemble-based versus expert-assisted approach to carbon price features selection

Ruszczak,

Rudnik

2023

Annals of Computer Science and Information Systems

View full text Add to dashboard Cite

The paper comments on two main issues. First, on a model for estimating the carbon price using multi-year market data. And second, on the consideration of two approaches to feature set exploitation. On the one hand, two ensemble machine-learning models with randomly selected feature sets are employed. On the other hand, a hybrid feature selection strategy follows domain expertise on which features should be explored. This minimizes the number of feature set combinations to be tested. The additional information for the predictions was the data from other commodity contracts, which could be easily introduced into the collection, as too many of them do not necessarily improve the estimates. The results of the experiments are promising: for the model based on SVR, the MAPE obtained was 2.09% and 5.6% for the following day and week price forecasts, respectively.

show abstract

Section: Literature Reviewmentioning

confidence: 97%

Ensemble-based versus expert-assisted approach to carbon price features selection

Ruszczak,

Rudnik

2023

Annals of Computer Science and Information Systems

View full text Add to dashboard Cite

show abstract

“…24 Through a component analysis of the carbon price, E et al used the least square support vector regression method to forecast the nonlinear fluctuation of the carbon price. 25 To predict the volatility of carbon price, Huang et al proposed a hybrid model by integrating GARCH and LSTM neural network for carbon price forecasting. 26 Although the effectiveness of these methods in the field of carbon price prediction has been verified, the predictive performance and reliability of a single model cannot be guaranteed.…”

Section: For Example Zhu Et Al Predicted the Development Trend Of Carbonmentioning

confidence: 99%

“…Based on a gated recurrent unit (GRU) neural network, Liu and Shen proposed a hybrid carbon price forecasting methodology 24 . Through a component analysis of the carbon price, E et al used the least square support vector regression method to forecast the nonlinear fluctuation of the carbon price 25 . To predict the volatility of carbon price, Huang et al proposed a hybrid model by integrating GARCH and LSTM neural network for carbon price forecasting 26 .…”

Section: Introductionmentioning

confidence: 99%

Prediction uncertainty and volatility for carbon price using an adaptive lower and upper bound estimation model

Yang

Zhiqiang

2023

Env Prog and Sustain Energy

View full text Add to dashboard Cite

The high volatility and uncertainty of carbon price have always been two major challenges in carbon price forecasting. To solve these two challenges, an adaptive lower-and upper-bound estimation (LUBE) model with improved variational mode decomposition (VMD) and PSO-based interval optimization strategy is proposed for interval prediction of carbon price. To validate effectiveness and superiority, the adaptive LUBE model and several competitive models, including the bootstrap model, delta model, and Bayesian model, were utilized for interval prediction of carbon prices of Beijing and Shanghai. Compared with other models, the adaptive LUBE model not only has excellent coverage but also has the narrowest interval width in both training set and test set. Therefore, the excellent comparison results show that the proposed model can obtain a more reliable and higher-quality prediction interval, which can be a novel and effective carbon prices forecasting tool for governments and enterprises.

show abstract

“…The EMD (Empirical Mode Decomposition, EMD) technology have proved to be an effective model to capture the non-linear and non-stationary characteristic of carbon price [1][2][3]. Conducted the EMD technology to decompose the carbon price, used the particle swarm optimization least squares support vector machine (PSO-LSSVM) model for out-of-sample forecasting, the results suggest the EMD-PSO-LSSVM model has forecasting superiority in Europe carbon price [4][5].To reduce the decomposition noise, the CEEMDAN and VMD technologies were used to perform primary and secondary decomposition of the original price, LSTM and ELM models were used for forecasting. The findings put that the CEEMDAN-VMD-LSTM and CEEMDAN-VMD-ELM models have substantially forecasting accuracy in China carbon market [6][7][8].…”

Section: Introductionmentioning

confidence: 99%

A Novel Hybrid Model for Forecasting China Carbon Price Using CEEMDAN and Extreme Learning Machine Optimized by Whale Algorithm

Ni,

Liew

2023

Advances in Transdisciplinary Engineering

View full text Add to dashboard Cite

The carbon market can provide economic incentives for manufacturing industry to reduce carbon emissions. This paper follows the idea of “primary decomposition- noise reduction-secondary decomposition- forecasting and integration”, the contribution is constructing a hybrid carbon price forecasting model using Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) and Extreme Learning Machine (ELM) optimized by the Whale Optimization Algorithm (WOA). The results conclude that, the CEEMDAN-type secondary decomposition hybrid models have high forecasting accuracy, the WOAELM-type models can effectively reduce the forecasting errors. Noteworthy, the forecasting errors RMSE, MAE and MAPE of the proposed CEEMDAN-SE-CEEMD-WOAELM model are 2.587, 2.04 and 0.108 respectively, that is the lowest in all the comparative models. The forecasting accuracy and reliability of the proposed model have been convinced. Those findings can provide valuable reference for manufacturing industry to reduce pollutant emissions and take low-carbon investment.

show abstract

A denoising carbon price forecasting method based on the integration of kernel independent component analysis and least squares support vector regression

Cited by 43 publications

References 46 publications

Ensemble-based versus expert-assisted approach to carbon price features selection

Ensemble-based versus expert-assisted approach to carbon price features selection

Prediction uncertainty and volatility for carbon price using an adaptive lower and upper bound estimation model

A Novel Hybrid Model for Forecasting China Carbon Price Using CEEMDAN and Extreme Learning Machine Optimized by Whale Algorithm

Contact Info

Product

Resources

About