基于Lasso回归的空气污染治理影响因素分析及治理绩效评价方法

Lasso Regression Based Analysis on Influence Factors of Air Pollution Control and Governance Performance Evaluation Method

利用Lasso回归方法,从投入产出的视角分析识别出对我国空气污染治理的主要影响因素,并测度空气污染治理主要输入指标对各空气质量指标的影响,进而提出了一种基于回归加权的空气污染治理绩效评价方法,改进了已有的空气污染指数(API)评价方法。与已有的综合评价方法相比,本文的方法通过回归分析设定权重,既保证了空气污染治理中各种投入指标在评价中的地位和作用,又保证了权重的客观性。最后以重庆市为例,演示了本文方法在实际分析中的应用过程,并对我国主要城市2017年的空气污染治理绩效进行了研究。

Based on the lasso regression method, we identify the main influencing factors on China’s air pollution control from the perspective of input-output, and measure the impact of the main input indexes of air pollution control on each air quality index. Then, we proposed an air pollution control performance evaluation method based on regression weighting, which modifies the existing air pollution index (API) evaluation method. Compared with the existing comprehensive evaluation methods, the method proposed in this paper sets the weight through regression analysis, which not only ensures the status and role of various input indicators in the evaluation of air pollution control, but also ensures the objectivity of the weight. Finally, taking Chongqing as an example, the paper demonstrates the application process of the proposed method in the practical analysis, and studies the air pollution control performance of the major cities in China in 2017.