Forecasting carbon price using empirical wavelet transform and gated recurrent unit neural network

Carbon price, to a certain extent, reflects the intensity of a national emission reduction target, whereas carbon price forecasting is the basis for improving crisis management competence and strengthening market enthusiasm. This paper advances a novel hybrid carbon price forecasting methodology con...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Carbon management Jg. 11; H. 1; S. 25 - 37
Hauptverfasser: Liu, Hui, Shen, Lei
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Taylor & Francis 02.01.2020
Taylor & Francis Group
Schlagworte:
administrative management
algorithms
autocorrelation
carbon
carbon markets
Carbon price forecasting
disaster recovery
empirical wavelet transform
fuzzy C-means clustering algorithm
gated recurrent unit neural network
prediction
wavelet
ISSN:1758-3004, 1758-3012, 1758-3012
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Carbon price, to a certain extent, reflects the intensity of a national emission reduction target, whereas carbon price forecasting is the basis for improving crisis management competence and strengthening market enthusiasm. This paper advances a novel hybrid carbon price forecasting methodology consisting of the empirical wavelet transform (EWT) and the gated recurrent unit (GRU) neural network. First, the carbon price data is decomposed through the EWT approach into the more stable and regular sub-components. These sub-components are divided into trend, low-frequency and high-frequency component using the fuzzy C-means clustering algorithm. Next, the lag order of different classes of components is determined as the input variables of the GRU model by the partial auto-correlation function method. Then, all values of each component predicted by the GRU method are aggregated to produce a final combined prediction result for the original carbon price. Finally, the EWT-GRU model is compared with the individual Autoregressive Integrated Moving Average (ARIMA), Back Propagation Neural Network (BPNN), GRU and EWT-BPNN models. The simulation results demonstrate that the proposed EWT-GRU combined forecasting model is superior to other models in terms of prediction effect, prediction accuracy, etc. They also confirm the validity and accuracy of the EWT-GRU model in carbon price prediction and show it deserves popularization.
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
ISSN:1758-3004
1758-3012
1758-3012
DOI:10.1080/17583004.2019.1686930