基于随机森林算法的中长期天然气需求影响因素分析

Analysis of factors influencing medium and long-term natural gas demand based on random forest algorithm

天然气在能源转型过程中发挥着重要作用,深入探究中长期天然气需求的关键影响因素至关重要。以中国和30个省为研究对象,对不同颗粒度中长期天然气需求影响因素进行系统梳理,收集处理近20年的历史数据,采用随机森林算法进行影响因素筛选和排序。结果表明:随机森林算法具有很好的适用性;从整体情况看,国内生产总值(GDP)、第二产业占比、第三产业占比、城镇化率、天然气消费占比、电力消费占比、城市天然气供气管道长度以及城市天然气用气人口等是相对重要的影响因素;不同省份影响因素则存在较大差异,各省在经济发展水平、资源禀赋、能源消费偏好、地理位置等方面的不同是导致差异出现的主要原因。因此,在开展天然气产业规划和政策制定时,建议因地制宜分类施策,促进天然气在新型能源体系建设中发挥更大作用。

Natural gas plays an important role in the process of energy transition,and it is crucial to deeply explore the key influencing factors of natural gas demand in the medium and long term.Taking China and 30 provinces as the research object,systematically sorted out the influencing factors of medium and long term natural gas demand with different granularity,the historical data from 2003 to 2022 were collected and processed,and the random forest algorithm was used to screen and sort the influencing factors.The results show that:The random forest algorithm is highly applicable;From the overall situation,GDP(gross domestic product),the proportion of secondary industry,the proportion of tertiary industry,the urbanization rate,the proportion of natural gas consumption,the proportion of electricity consumption,the length of urban natural gas supply pipeline and the population of urban natural gas use are relatively important influencing factors;The influencing factors are different in different provinces,There are great differences in influencing factors among different provinces,which are mainly caused by characteristics in economic development level,resource endowment,energy consumption preference and geographical location.Therefore,when carrying out natural gas industry planning and policy formulation,it is recommended to give classified policies according to local conditions,so as to promote natural gas to play a greater role in the construction of a new energy system.