An interval-valued carbon price forecasting method based on web search data and social media sentiment

Area of Science:

• Environmental Economics
• Computational Finance
• Data Science

Background:

• Accurate carbon price prediction is vital for carbon trading markets.
• Existing methods often overlook online data and rely on point predictions, limiting forecasting accuracy.
• Challenges remain in precisely forecasting carbon prices due to data limitations.

Purpose of the Study:

• To propose an interval-valued carbon price forecasting method incorporating web search data and social media sentiment.
• To enhance prediction performance by synthesizing diverse data sources.
• To improve the accuracy and reliability of carbon price forecasting.

Main Methods:

• Collected and synthesized web search data and social media sentiment.
• Applied Principal Component Analysis (PCA) for high-dimensional web search data preprocessing.
• Utilized BosonNLP for social media sentiment quantification.
• Employed Variational Mode Decomposition (VMD) on carbon prices and online data.
• Used Particle Swarm Optimization Support Vector Regression (PSO-SVR) for sub-mode prediction and aggregation.

Main Results:

• Web search data and social media sentiment significantly enhance the predictive accuracy of interval-valued carbon prices.
• The proposed VMD-PSO-SVR model demonstrated superior performance in interval-valued forecasting accuracy and reliability.
• Case studies in Guangdong and Hubei provinces validated the effectiveness of the integrated approach.

Conclusions:

• Integrating online data sources like web search trends and social media sentiment is effective for improving carbon price forecasting.
• The VMD-PSO-SVR method provides a robust framework for accurate and reliable interval-valued carbon price predictions.
• This research offers valuable insights for carbon trading market participants and policymakers.