Table 1.
Economic and social indicators for allocating urban carbon emissions.
Fig 1.
The geographic location and city names of 16 cities with data from 2000 to 2017.
Areas of cities are shaded by colours while the Yangtze River is indicated by lines. Data source are retrieved from Ministry of Natural Resources of the People’s Republic of China (grant number: NO. GS(2020)4621; http://bzdt.ch.mnr.gov.cn/index.html).
Table 2.
Ten influencing factors and the short names used.
Fig 2.
Curve of the mean squared errors resulting from RF varying with the number of decision trees (NT).
Fig 3.
The top 30 normalized Gini impurities resulting from RF.
Fig 4.
Correlation between 10 influencing factors and carbon emissions during 2000–2017 in 16 cities.
Each row represents a particular city, while each column represents a particular influencing factor. Colours in each pixel represent correlation, while labels *, **, *** separately indicate statistical significance in 90%, 99%, 99.9% confidence intervals.
Fig 5.
The Gini impurity of 10 impact factors for 18 cities.
The plot is organized following the order of three city agglomerations introduced in Section 3.2 (Groups 1–3). Each panel represents an individual city, labelled in the title of each panel.
Fig 6.
The difference (error) rate between predicted and real CDE in each city during 2012 to 2017.
Each row represents a particular city, while each column represents a particular year.
Table 3.
Mean absolute difference of 16 cities’ CDE obtained by RF cross–validation.
The mean absolute difference for each city is calculated by temporally averaging all differences between predicted and real CDE (shown in Fig 5) for the corresponding city.