基于决策树不同模型下哈密市PM_2.5预测分析
PM_2.5 Predictive Analysis in Hami City Based on Different Decision Tree Models

为提升哈密市区域PM_2.5浓度预测模型的准确度与时效性，研究基于机器学习范式，遴选决策树(Decision Tree, DT)、随机森林(Random Forest, RF)及梯度提升决策树(Gradient Boosting Decision Tree, GBDT)三类典型集成学习算法，针对巴里坤站、伊州区站与伊吾站三处典型环境监测站点，开展多污染情景下的预测效能对比分析。通过构建污染特征矩阵与时空耦合数据集，系统考察不同气象–排放复合污染场景中模型的动态响应特性。
To improve the accuracy and timeliness of the PM_2.5 concentration prediction model in Hami City, this study, based on the machine learning paradigm, selects three typical ensemble learning algorithms: Decision Tree (DT), Random Forest (RF), and Gradient Boosting Decision Tree (GBDT). It conducts a comparative analysis of prediction performance under multiple pollution scenarios for three typical environmental monitoring stations, Balikun Station, Yizhou Station, and Yiwu Station. By constructing a pollution feature matrix and a spatiotemporal coupled dataset, the dynamic response characteristics of the models in different meteorological-emission composite pollution scenarios are systematically examined.