Study on the influencing factors of low carbon dioxide travel based on machine learning

The aim of this study was to identify key factors affecting urban low carbon dioxide travel levels and analyse their impact mechanisms. Using Wuhan in China as a case study, a method to identify residents' travel modes through mobile signalling data and calculate regional low-carbon travel (LCT) levels was developed. Interpretable machine learning algorithms were applied to analyse the factors and mechanisms driving differences in LCT levels at a 1 km × 1 km scale. The results show that land use density and land use diversity are the most significant influences, followed by building density, surface roughness and public transportation provision. Among various points of interest considered, the density of parking lots and residential areas was found to have a notable impact. The relationship between the indicator values of influencing factors and urban LCT levels was also examined. Key points corresponding to marginal effects on LCT levels were identified, providing insights for developing precise travel strategies in urban planning.