A Hybrid LSTM- Random Forest Model for Carbon Emission Prediction with Spatiotemporal Features in the Yangtze River Delta Region

Accurate carbon emission prediction is crucial for addressing climate change and promoting sustainable development. Traditional methods, primarily based on statistical regression models, often fail to capture the complex nonlinear relationships in carbon emission data. This study proposes a hybrid prediction framework integrating Long Short-Term Memory (LSTM) and Random Forest (RF) models. The LSTM component captures temporal dependencies in carbon emissions, while the RF model incorporates spatial features such as economic activity and energy consumption to enhance prediction accuracy. Experimental validation using carbon emission data and economic indicators from the Yangtze River Delta region demonstrates that the LSTM-RF hybrid model significantly outperforms traditional statistical regression methods in processing regional carbon emission data. Future research will focus on optimizing the model structure and integrating additional external factors (e.g., policy changes, technological advancements) to further enhance predictive capabilities. This methodology not only applies to carbon emission prediction but can also be extended to other environmental monitoring domains, providing technical support for global climate change mitigation and sustainable development.