基于Lasso-GRNN神经网络模型的北京市物流业碳排放量预测
Prediction of Carbon Emissions from the Logistics Industry in Beijing Based on Lasso-GRNN Neural Network Model

在以“碳达峰、碳中和”为发展目标的背景下，本文构建了一个组合预测方法——Lasso-GRNN神经网络模型对北京市物流业的碳排放量进行分析预测。首先测算2000~2021年北京市物流业碳排放量，选取地区生产总值、物流业产值、贸易出口总额等20项指标变量作为北京市物流业碳排放量的影响因素，利用Lasso回归模型确定影响北京市物流业碳排放量的关键变量，在此基础上将筛选出的各指标值作为GRNN神经网络的输入变量，构建Lasso-GRNN神经网络模型对碳排放量进行预测。研究结果表明，Lasso-GRNN神经网络模型的预测效果明显优于PLS-GRNN和PCA-GRNN组合预测模型，该模型误差更小、精度更高，更适合用于碳排放量及其相关指标的预测。
In the context of “carbon peak and carbon neutrality” as the development goal, this paper constructs a combined prediction method—Lasso-GRNN neural network model, to analyze and predict the carbon emissions of logistics industry in Beijing. Firstly, the carbon emissions of logistics industry in Beijing from 2000 to 2021 are calculated, and 20 index variables such as gross regional product, output value of logistics industry and total trade exports are selected as the influencing factors of carbon emissions of logistics industry in Beijing. Lasso regression model is adopted to determine the key variables affecting carbon emissions of logistics industry in Beijing. On this basis, the selected index values are taken as the input variables of GRNN neural network, and the Lasso-GRNN neural network model is constructed to predict the carbon emission. The results show that the prediction effect of the Lasso-GRNN neural network model is significantly better than that of PLS-GRNN and PCA-GRNN combined prediction models, which has smaller error and higher precision, and is more suitable for the prediction of carbon emission and its related indicators.