Related Experiment Videos
Forecasting Natural Gas Prices Using Wavelets, Time Series, and Artificial Neural Networks
1Department of Natural Resources and Environmental Engineering, Hanyang University, Seoul, Korea.
Plos One
|November 6, 2015
Summary
Forecasting natural gas prices is crucial due to its decoupling from crude oil. Hybrid models combining wavelet decomposition with time series methods offer superior accuracy, efficiently handling boundary issues for reliable predictions.
Area of Science:
- Energy Economics
- Time Series Analysis
- Computational Finance
Background:
- The unconventional gas revolution has weakened the link between natural gas and crude oil prices.
- Accurate natural gas price forecasting is essential for market participants.
- Traditional forecasting models struggle with the complexities of energy markets.
Purpose of the Study:
- To develop and evaluate novel hybrid models for natural gas price forecasting.
- To investigate the impact of wavelet decomposition, including boundary problem handling, on forecasting accuracy.
- To compare the performance of different hybrid model combinations.
Main Methods:
- Hybrid models integrating wavelet decomposition (approximation and detail components) with Autoregressive Integrated Moving Average (ARIMA) and Generalized Autoregressive Conditional Heteroskedasticity (GARCH) models.
- Artificial Neural Networks (ANNs) were also incorporated into hybrid structures.
- Comparison of forecasting results with and without addressing the wavelet decomposition boundary problem.
Main Results:
- Wavelet-hybrid models demonstrated superior performance in all tested scenarios.
- Effectively handling the boundary problem in wavelet decomposition improved forecasting accuracy.
- Utilizing only the approximation component of wavelet decomposition offered acceptable forecasting efficiency.
Conclusions:
- Hybrid models incorporating wavelet decomposition provide a robust approach to natural gas price forecasting.
- Addressing the boundary problem is critical for optimizing wavelet-based forecasting.
- Forecasting using the approximation component alone is a viable option balancing accuracy and efficiency.
