ScienceGuardians

ScienceGuardians

Did You Know?

ScienceGuardians holds all parties accountable

A Decomposition-Integration Framework of Carbon Price Forecasting Based on Econometrics and Machine Learning Methods

Authors: Zhehao Huang,Benhuan Nie,Yuqiao Lan,Changhong Zhang
Publisher: MDPI AG
Publish date: 2025-1-30
ISSN: 2227-7390 DOI: 10.3390/math13030464
View on Publisher's Website
Up
0
Down
::

The study has several critical issues that require attention. The application of GARCH to potentially non-stationary decomposed components is methodologically flawed, and the CEEMDAN parameter choices lack proper justification. Additionally, the use of GWO for hyperparameter tuning is not benchmarked against standard optimization methods, making its effectiveness uncertain. I invite the authors to address these concerns to strengthen the study’s validity and reliability.

All Replies

Viewing 1 replies (of 1 total)

2 weeks, 1 day ago

The concerns raised are valid, and the study presents the following approach: Only stationary IMFs, confirmed via ADF tests, are modeled with GARCH to ensure correctness. CEEMDAN parameters are selected based on empirical benchmarks and validated using complexity measures like Lempel–Ziv and dispersion entropy. GWO is chosen for hyperparameter tuning due to its efficiency in avoiding local optima, with performance tested against standard configurations. These choices aim to enhance accuracy and robustness, but if there are aspects that could be improved or alternative perspectives to consider, further discussion would be valuable.

Viewing 1 replies (of 1 total)

  • You must be logged in to reply to this topic.