中国碳强度关键影响因子的机器学习识别及其演进

doi:10.11821/dlxb201912012

地理学报 ›› 2019, Vol. 74 ›› Issue (12): 2592-2603.doi: 10.11821/dlxb201912012

• 资源环境与可持续发展 • 上一篇下一篇

中国碳强度关键影响因子的机器学习识别及其演进

刘卫东^1,², 唐志鹏^1,²(), 夏炎³, 韩梦瑶¹, 姜宛贝¹

^1. 中国科学院地理科学与资源研究所中国科学院区域可持续发展分析与模拟重点实验室,北京 100101
^2. 中国科学院大学资源与环境学院,北京 100049
^3. 中国科学院科技战略咨询研究院,北京 100190

收稿日期:2018-05-08 修回日期:2019-10-17 出版日期:2019-12-25 发布日期:2019-12-25
作者简介:刘卫东(1967-), 男, 河北隆化人, 研究员, 中国地理学会会员(S110001202M), 主要从事经济地理和区域发展研究。E-mail:liuwd@igsnrr.ac.cn
基金资助:
国家重点基础研究发展计划(2016YFA0602804);国家自然科学基金项目(41771135)

Identifying the key factors influencing Chinese carbon intensity using machine learning, the random forest algorithm, and evolutionary analysis

LIU Weidong^1,², TANG Zhipeng^1,²(), XIA Yan³, HAN Mengyao¹, JIANG Wanbei¹

^1. Key Laboratory of Regional Sustainable Development Modeling, Institute of Geographic Sciences and Natural Resources Research, CAS, Beijing 100101, China
^2. College of Resources and Environment, University of Chinese Academy of Sciences, Beijing 100049, China
^3. Institute of Science and Development, CAS, Beijing 100190, China

Received:2018-05-08 Revised:2019-10-17 Published:2019-12-25 Online:2019-12-25
Supported by:
The National Key Research and Development Program of China(2016YFA0602804);National Natural Science Foundation of China(41771135)

摘要/Abstract

摘要：

碳强度影响因子数量众多,通过在众多因子中评估其重要性以识别出关键影响因子进而解析碳强度关键因子的变化规律,是中国2030年碳强度能否实现比2005年下降60%~65%目标的科学基础。传统的回归分析方法对于评估众多因子的重要性存在多重共线性等问题,而机器学习处理海量数据则具有较好的稳健性等优点。本文从能源结构、产业结构、技术进步和居民消费等方面选取了56个中国碳强度影响因子指标,采用随机森林算法基于信息熵评估了1980-2014年逐年各项因子的重要性,通过指标数量与信息熵的对应关系统一筛选出每年重要性最大的前22个指标作为相应年度关键影响因子,最终依据关键影响因子的变化趋势划分了3个阶段作了演进分析。结果发现：1980-1991年,碳强度的关键因子主要以高耗能产业规模及占比、化石能源占比和技术进步为主;1992-2007年,中国经济进入快车道增长时期,服务业占比和化石能源价格对碳强度的影响作用开始显现,居民传统消费的影响作用在增大;2008年全球金融危机后,中国进入经济结构深化调整时期,节能减排力度大大增强,新能源占比和居民新兴消费的影响作用迅速显现。为实现2030年碳强度下降60%~65%目标,优化能源结构和产业结构,促进技术进步,提倡绿色消费,强化政策调控是未来需要采取的主要措施。

关键词: 机器学习, 随机森林, 碳强度, 关键影响因子, 中国

Abstract:

As the Chinese government ratified the Paris Climate Agreement in 2016, the goal of reducing carbon dioxide emissions per unit of gross domestic product (carbon intensity) from 60% to 65% of 2005 levels must now be achieved by 2030. However, as numerous factors influence Chinese carbon intensity, it is key to assess their relative importance in order to determine which are most important. As traditional methods are inadequate for identifying key factors from a range acting simultaneously, machine learning is applied in this research. The random forest (RF) algorithm based on decision tree theory was proposed by Breiman (2001); this algorithm is one of the most appropriate because it is insensitive to multicollinearity, robust to missing and unbalanced data, and provides reasonable predictive results. We therefore identified the key factors influencing Chinese carbon intensity using the RF algorithm and analyzed their evolution between 1980 and 2014. The results of this analysis reveal that dominant factors include the scale and proportion of energy-intensive industries as well as fossil energy proportion and technical progress between 1980 and 1991. As the Chinese economy developed rapidly between 1992 and 2007, effects on carbon intensity were enhanced by service industry proportion and the fossil fuel price such that the influence of traditional residential consumption also increased. The Chinese economy then entered a period of deep structural adjustment subsequent to the 2008 global financial crisis; energy-saving emission reductions were greatly enhanced over this period and effects on carbon intensity were also rapidly boosted by the increasing availability of new energy and its residential consumption. Optimization of energy and industrial structures, promotion of technical progress, green consumption, and the reduction and management of emissions will be key to cutting future carbon intensity levels within China. These approaches will all help to achieve the 2030 goal of reducing carbon emission intensity from 60% to 65% of 2005 levels.

Key words: machine learning, random forest, carbon intensity, key factor, China

刘卫东, 唐志鹏, 夏炎, 韩梦瑶, 姜宛贝. 中国碳强度关键影响因子的机器学习识别及其演进[J]. 地理学报, 2019, 74(12): 2592-2603.

LIU Weidong, TANG Zhipeng, XIA Yan, HAN Mengyao, JIANG Wanbei. Identifying the key factors influencing Chinese carbon intensity using machine learning, the random forest algorithm, and evolutionary analysis[J]. Acta Geographica Sinica, 2019, 74(12): 2592-2603.

图/表 5

表1

中国碳强度影响因子的类别和指标

类别	具体因子指标(单位)	类别	具体因子指标(单位)
化石能源占比	煤炭占比(%)	高能耗产业规模及占比	纯碱(万t)
	石油占比(%)		烧碱(万t)
	天然气占比(%)		乙烯(万t)
化石能源价格	煤炭行业出厂价格指数		合成氨(万t)
化石能源价格	石油和天然气行业出厂价格指数		水泥(万t)
非化石能源占比	水电占比(%)		平板玻璃(万重量箱)
非化石能源占比	沼气占比(%)		粗钢(万t)
新能源占比	风电占比(%)		成品钢(万t)
	核能占比(%)		建筑业占比(%)
	光伏占比(%)		交通运输、仓储和邮政业占比(%)
	光热占比(%)	技术进步	全员劳动生产率(万元/人)
	地热占比(%)		发电及电站供热加工转换效率(%)
	液体生物燃料占比(%)		炼焦加工转换效率(%)
服务业占比	金融业占比(%)		炼油加工转换效率(%)
	信息传输、计算机服务和软件业占比(%)		发电标准煤耗(克/千瓦时)
	教育业占比(%)		供电标准煤耗(克/千瓦时)
	卫生和社会保障社会福利占比(%)		发电厂线路损失率(%)
	文化体育娱乐业占比(%)		粗钢行业单位综合耗能(吨标煤/t)
	科学研究、技术服务和地质勘查业占比(%)		水泥行业单位综合耗能(千克标煤/t)
居民传统消费	城镇每百户私人汽车(辆)		乙烯行业单位综合耗能(吨标煤/t)
	城镇每百户摩托车(辆)		合成氨行业单位综合耗能(吨标煤/t)
	农村每百户摩托车(辆)		科技拨款占财政总支出的占比(%)
	城镇每百户电冰箱(台)		公共建筑单位面积耗能(千克标煤/m²)
	农村每百户电冰箱(台)		城镇居住建筑单位面积耗能(千克标煤/m²)
	城镇每百户电视机(台)		农村居住建筑单位面积耗能(千克标煤/m²)
	农村每百户电视机(台)	居民新兴消费	互联网宽带接入用户(万户)
	城镇每百户洗衣机(台)	居民新兴消费	高铁营业里程(km)
	农村每百户洗衣机(台)		电子商务交易额(万亿)

表1

表2

图1

表3

图2

参考文献 30

[1]	Zhang Youguo . Economic development pattern change impact on China's carbon intensity. Economic Research Journal, 2010(4):120-133.
	[ 张友国 . 经济发展方式变化对中国碳排放强度的影响. 经济研究, 2010(4):120-133.]
[2]	Stern D, Jotzo F . How ambitious are China and India's emissions intensity targets? Energy Policy, 2010,38(11):6776-6783. doi: 10.1016/j.enpol.2010.06.049
[3]	Yuan J, Hou Y, Xu M . China's 2020 carbon intensity target: Consistency, implementations, and policy implications. Renewable and Sustainable Energy Reviews, 2012,16(7):4970-4981. doi: 10.1016/j.rser.2012.03.065
[4]	Wang Shaohua, Yu Weiyang . Sensitivity analysis of primary energy consumption structural change and carbon intensity. Resources Science, 2013,35(7):1438-1446.
	[ 王韶华, 于维洋 . 一次能源消费结构变动对碳强度影响的灵敏度分析. 资源科学, 2013,35(7):1438-1446.]
[5]	Peng Xu, Cui Herui . Research on the effects of energy structure adjustment in china on carbon intensity. Journal of Dalian University of Technology (Social Sciences), 2016,37(1):11-16.
	[ 彭旭, 崔和瑞 . 中国能源结构调整对碳强度的影响研究. 大连理工大学学报(社会科学版), 2016,37(1):11-16.]
[6]	Li H, Wang L, Shen L , et al. Study of the potential of low carbon energy development and its contribution to realize the reduction target of carbon intensity in China. Energy Policy, 2012,41:393-401. doi: 10.1016/j.enpol.2011.10.061
[7]	Fan Y, Liu L C, Wu G , et al. Changes in carbon intensity in China: Empirical Findings from 1980-2003. Ecological Economics. 2007,62(3/4):683-691. doi: 10.1016/j.ecolecon.2006.08.016
[8]	Zhang Y G . Structural decomposition analysis of sources of decarbonizing economic development in China: 1992-2006. Ecological Economics, 2009,68(8/9):2399-2405. doi: 10.1016/j.ecolecon.2009.03.014
[9]	Feng Yan, Zhu Lingyun, Zhang Dahong . Spatial and econometric analysis of effect of industrial structure adjustment on carbon intensity in China. Soft Science, 2017,31(7):11-15.
	[ 冯彦, 祝凌云, 张大红 . 中国产业结构调整对碳强度影响的空间计量研究. 软科学, 2017,31(7):11-15.]
[10]	Xu Hui, Wang Hui . The impact of industrial structure adjustment on China's carbon intensity goal: The outlook of 2020. Science and Technology Management Research, 2016,36(13):232-236.
	[ 徐辉, 王慧 . 产业结构调整对中国碳强度目标实现的影响: 展望2020年. 科技管理研究, 2016,36(13):232-236.]
[11]	Yan Zheming, Deng Xiaolan, Yang Zhiming . The impact of heterogeneous technological innovation on carbon intensity: A global evidence based on patent statistics. Journal of Beijing Institute of Technology (Social Sciences Edition), 2017,19(1):20-27.
	[ 鄢哲明, 邓晓兰, 杨志明 . 异质性技术创新对碳强度的影响: 基于全球专利数据. 北京理工大学学报(社会科学版), 2017,19(1):20-27.]
[12]	Huang Jie, Ding gang . A research on the threshold effect of the impact of technical progress on carbon intensity. Science & Technology Progress and Policy, 2014,31(18):22-26. doi: 10.1111/cxo.13030 pmid: 31852027
	[ 黄杰, 丁刚 . 技术进步对碳强度影响的门槛效应研究. 科技进步与对策, 2014,31(18):22-26.] doi: 10.1111/cxo.13030 pmid: 31852027
[13]	Voigt S, Cian E, Schymura M , et al. Energy intensity developments in 40 major economies: Structural change or technology improvement? Energy Economics, 2014,41:47-62. doi: 10.1016/j.eneco.2013.10.015
[14]	Dong Mei, Xu Zhangyong, Li Cunfang . The impact of carbon intensity restriction on welfare of urban and rural residents: An analysis based on CGE model. China Population, Resources and Environment, 2018,28(2):94-105.
	[ 董梅, 徐璋勇, 李存芳 . 碳强度约束对城乡居民福利水平的影响: 基于CGE模型的分析. 中国人口·资源与环境, 2018,28(2):94-105.]
[15]	Tong Jinping, Chen Guodong, Yang Zuying , et al. Threshold effects of household consumption level on residential carbon emissions. Journal of Arid Land Resources and Environment, 2017,31(1):38-43.
	[ 佟金萍, 陈国栋, 杨足膺 , 等. 居民消费水平对生活碳排放的门槛效应研究. 干旱区资源与环境, 2017,31(1):38-43.]
[16]	Fan J, Liao H, Liang Q M , et al. Residential carbon emission evolutions in urban-rural divided China: An end-use and behavior analysis. Applied Energy, 2013,101:323-332. doi: 10.1016/j.apenergy.2012.01.020
[17]	Zhu Zhiyuan, Miao Jianjun, Cui Wei . Analysis of carbon emission efficiency for urban construction land intensive use. Areal Research and Development, 2016,35(3):98-103.
	[ 朱志远, 苗建军, 崔玮 . 城市建设用地集约利用的碳排放效率分析. 地域研究与开发, 2016,35(3):98-103.]
[18]	You Heyuan, Wu Cifang . Analysis of carbon emission efficiency and optimization of low carbon for agricultural land intensive use. Transactions of the Chinese Society of Agricultural Engineering, 2014,30(2):224-234.
	[ 游和远, 吴次芳 . 农地集约利用的碳排放效率分析与低碳优化. 农业工程学报, 2014,30(2):224-234.]
[19]	Zhang Miao, Gan Chenlin, Chen Yinrong , et al. Carbon emission efficiency and optimization of low carbon for construction land development intensity in China according to provincial panel data. Resources Science, 2016,38(2):265-275. doi: 10.18402/resci.2016.02.09
	[ 张苗, 甘臣林, 陈银蓉 , 等. 中国城市建设用地开发强度的碳排放效率分析与低碳优化. 资源科学, 2016,38(2):265-275.] doi: 10.18402/resci.2016.02.09
[20]	Zhu Feifei, Zhou Lin, Mo Leping , et al. A scientific study of establishing statistical index system. Statistical Theory and Practice, 2015(11):8-12.
	[ 朱飞飞, 周琳, 莫乐平 , 等. 统计指标体系设置的科学性研究. 统计科学与实践, 2015(11):8-12.]
[21]	Greening L A, Ting M, Krackler T J . Effects of changes in residential end-uses and behavior on aggregate carbon intensity: Comparison of 10 OECD countries for the period 1970 through 1993. Energy Economics, 2001,23(2):153-178. doi: 10.1016/S0140-9883(00)00059-1
[22]	Ebohon O J, IKeme A J . Decomposition analysis of CO₂ emission intensity between oil-producing and non-oil-producing sub-Saharan African countries. Energy Policy, 2006,34(18):3599-3611. doi: 10.1016/j.enpol.2004.10.012
[23]	Gingrich S, Kušková P, Steinberger J K . Long-term changes in CO₂ emissions in Austria and Czechoslovakia-Identifying the drivers of environmental pressures. Energy Policy, 2011,39(2):535-543. doi: 10.1016/j.enpol.2010.10.006
[24]	Kohavi R, Foster P . Special issue on application of machine learning and the knowledge discovery process. Journal of Machine learning, 1998,30(2):271-274. doi: 10.1023/A:1017181826899
[25]	Chen Kai, Zhu Yu . A summary of machine learning and related algorithms. Statistics and Information Forum, 2007,22(5):105-112.
	[ 陈凯, 朱钰 . 机器学习及其相关算法综述. 统计与信息论坛, 2007,22(5):105-112.]
[26]	Iverson L R, Prasad A M, Matthews S N , et al. Estimating potential habitat for 134 eastern US tree species under six climate scenarios. Forest Ecology and Management, 2008,254(3):390-406. doi: 10.1016/j.foreco.2007.07.023
[27]	Breiman L . Random forests. Machine Learning, 2001,45(1):5-32. doi: 10.1023/A:1010933404324
[28]	Breiman L . Bagging predictors. Machine Learning, 1996,24(2):123-140. doi: 10.1016/j.envint.2019.104934 pmid: 31229871
[29]	Cao Zhengfeng . Study on optimization of random forests algorithm[D]. Beijing: Capital University of Economics and Business, 2014. doi: 10.1097/PCC.0000000000002121 pmid: 31805020
	[ 曹正风 . 随机森林算法优化研究[D]. 北京: 首都经济贸易大学, 2014.] doi: 10.1097/PCC.0000000000002121 pmid: 31805020
[30]	Breiman L, Friedman J, Stone C J , et al. Classification and Regression Trees. UK: Chapman & Hall/ CRC, 1984.

碳强度因子指标数量	1	2	3	4	5	6	7	8
Gini系数减小值	0.784	0.655	0.525	0.449	0.407	0.390	0.380	0.371
碳强度因子指标数量	9	10	11	12	13	14	15	16
Gini系数减小值	0.360	0.350	0.342	0.335	0.328	0.321	0.310	0.302
碳强度因子指标数量	17	18	19	20	21	22	23	24
Gini系数减小值	0.292	0.287	0.278	0.268	0.263	0.250	0.229	0.201
碳强度因子指标数量	25	26	27	28	29	30	31	32
Gini系数减小值	0.194	0.189	0.184	0.180	0.175	0.170	0.166	0.160
碳强度因子指标数量	33	34	35	36	37	38	39	40
Gini系数减小值	0.154	0.147	0.142	0.135	0.127	0.115	0.106	0.085
碳强度因子指标数量	41	42	43	44	45	46	47	48
Gini系数减小值	0.078	0.072	0.067	0.060	0.054	0.046	0.037	0.032
碳强度因子指标数量	49	50	51	52	53	54	55	56
Gini系数减小值	0.028	0.027	0.024	0.021	0.018	0.014	0.010	0.005

类别\年份	1980	1981	…	2000	…	2010	…	2013	2014
化石能源占比	3	3	…	0	…	1	…	2	1
化石能源价格	0	0	…	0	…	0	…	2	2
可再生能源(水电和沼气)占比	0	0	…	0	…	0	…	1	1
新能源占比	0	0	…	0	…	1	…	1	2
高耗能产业规模及占比	8	8	…	7	…	6	…	4	4
服务业占比	0	0	…	1	…	2	…	2	2
技术进步	8	8	…	6	…	4	…	3	5
居民传统消费	3	3	…	8	…	6	…	5	4
居民新兴消费	0	0	…	0	…	2	…	2	1
合计	22	22	…	22	…	22	…	22	22

中国碳强度关键影响因子的机器学习识别及其演进

Identifying the key factors influencing Chinese carbon intensity using machine learning, the random forest algorithm, and evolutionary analysis

RichHTML

PDF (PC)

赞

可视化

摘要/Abstract

引用本文

使用本文

图/表 5

参考文献 30

相关文章 15

Metrics

本文评价

[1]	朱晓华, 张燕, 朱媛媛. 大食物观视角下中国耕地保护的分区调控与区间协同[J]. 地理学报, 2023, 78(9): 2147-2162.
[2]	郑瑜晗, 龙花楼. 中国城乡融合发展测度评价及其时空格局[J]. 地理学报, 2023, 78(8): 1869-1887.
[3]	张溢堃, 王永生. 中国省域城乡要素流动测度方法与时空特征[J]. 地理学报, 2023, 78(8): 1888-1903.
[4]	夏添, 夏迎, 刘晓宇, 孙久文. 中国区域经济发展与政策体系演化——基于动力视角的三维分析框架[J]. 地理学报, 2023, 78(8): 1904-1919.
[5]	段德忠, 金红. 中国城市绿色技术扩散的时空过程与形成机制[J]. 地理学报, 2023, 78(8): 2001-2018.
[6]	冯晓华, 邱思远. 中国城市国内国际双循环测度及影响因素[J]. 地理学报, 2023, 78(8): 2019-2040.
[7]	柯文前, 肖宝玉, 林李月, 朱宇, 王焱. 中国省际城乡流动人口空间格局演变及与区域经济发展的关系[J]. 地理学报, 2023, 78(8): 2041-2057.
[8]	王俊松, 刘芷晴. 中国专利代理服务网络的时空演化格局及影响因素[J]. 地理学报, 2023, 78(8): 2058-2073.
[9]	戈大专. 新时代中国乡村空间特征及其多尺度治理[J]. 地理学报, 2023, 78(8): 1849-1868.
[10]	王光谦, 张宇, 谢笛, 钟德钰. 中国绿水格局及其战略意义[J]. 地理学报, 2023, 78(7): 1641-1658.
[11]	刘海龙, 张丽萍, 王炜桥, 张羽, 王争磊, 唐飞, 郭晓佳. 中国省际边界区县域城镇化空间格局及影响因素[J]. 地理学报, 2023, 78(6): 1408-1426.
[12]	周榕, 石磊, 庄汝龙. 中国旅居养老空间发展模式研究[J]. 地理学报, 2023, 78(6): 1553-1572.
[13]	冉钊, 高建华, 杨捷, 李晨阳. 中国商业健身资源集聚的格局演化及其形成机理[J]. 地理学报, 2023, 78(6): 1573-1590.
[14]	王强, 许有鹏, 杨龙, 童建, 林芷欣, 高斌. 太湖平原地区洪水特征变化及驱动机理[J]. 地理学报, 2023, 78(5): 1088-1103.
[15]	张明羽, 张正勇, 刘琳, 张雪莹, 康紫薇, 陈泓瑾, 高煜, 王统霞, 余凤臣. 中国天山山体效应评估及空间分异归因[J]. 地理学报, 2023, 78(5): 1254-1270.