Greenhouse Gases Time Series Forecasting using an Improved Hybrid Model Based on Discrete Wavelet Decomposition and Support Vector Regression (DWT-SVR)

Forecasting greenhouse gas (GHG) emissions helps countries plan future GHG reduction activities. Several machine learning-based techniques were applied for this purpose. This paper explores the potential and performance of a hybrid model that combines the discrete wavelet transform and support vector regression (DWT-SVR model) for GHG Time Series Forecasting. The model is constructed and applied to available GHG concentration data collected in California, USA. In addition, the model’s accuracy is compared to the traditional SVR. The experimental results show that the hybrid model DWT-SVR outperforms traditional SVR models, and the proposed approach has the highest forecasting accuracy. As a result, relying on the considered model potentially leads to better planning for future related activities.