【摘要】 文章提出了一种基于改进UMAP和强化学习优化随机森林的碳价格预测模型。改进后的UMAP结合时间信息和动态超参数调整，提高了时间相关性特征的保留能力，使RMSE降低了50.2%。随机森林模型采用强化学习动态调整树深度，避免过拟合与欠拟合，使RMSE降低了31.7%。实验结果表明，该方法在预测精度和效率方面优于传统方法，提升了预测模型的稳定性，可以为碳市场交易策略和政策制定提供支持，推动低碳经济发展。更多还原

【Abstract】 In this paper,a carbon price prediction model based on improved UMAP and reinforcement learning optimized random forest is proposed.The improved UMAP combines temporal information and dynamic hyperparameter tuning to improve the retention of time-dependent features,which reduces the RMSE by 50.2%.The random forest model,on the other hand,uses reinforcement learning to dynamically adjust the tree depth to avoid overfitting and underfitting,which reduces the RMSE by 31.7%.The experimental results show that the method outperforms traditional methods in terms of prediction accuracy and efficiency,improves the stability of the prediction model,and can provide support for carbon market trading strategies and policy formulation to promote the development of low-carbon economy.更多还原