Vehicle CO2 Emission Prediction Using MAWRF-AESL Model

Over the past few decades, rising CO₂ emissions from vehicles have become a major environmental concern, necessitating accurate predictive models for sustainable solutions. This study proposes a novel MAWRF-AESL hybrid model, integrating Modified Adaptive Weighted Random Forest (MAWRF) with Attention-Enhanced Sequential Learning (AESL) to enhance prediction accuracy. The MAWRF model efficiently extracts relevant features using an adaptive weighted tree-based approach, while AESL refines the predictions through LSTM networks with an attention mechanism. By sequentially processing the data—first with MAWRF and then AESL—the model achieves superior accuracy, significantly improving mean squared error (MSE) and R² scores compared to conventional methods. The results demonstrate near 99% prediction accuracy, making the model highly reliable for real-world CO₂ emission assessments. Future work will focus on expanding the dataset, optimizing hyperparameters, and integrating real-time data streams for enhanced adaptability and precision. Key Words: Vehicle emissions, CO2 prediction models, Machine learning, Predictive modelling, Hybrid prediction frameworks