基于深度学习的通勤碳排放家庭因素研究

Study on Family Characteristics in Commuting Carbon Emissions Based on Deep Learning

摘要: 通勤对于全球气候挑战至关重要。本研究旨在研究家庭对通勤行为的影响。通过整合常见的家庭特征，并采用TabNet模型对温岭市的5663份居民出行日常数据进行分析，发现家庭特征在大多数出行者的通勤碳排放中扮演着相较于个人特征更为重要的角色。其中，家庭中的主要出行方式、交通工具拥有量以及家庭成员的平均出行次数在这一影响中起着主导作用。研究结果还显示，这一影响在女性、未成年、高龄和低收入人群中并不明显，而在男性、中青年和高收入人群中则呈现出不同程度的相似性。基于以上结论，本研究建议低碳政策的制定应针对不同人群的家庭情况进行有针对性的制定，并可以根据家庭情况的变化反过来评估低碳减排政策的实施效果。

Abstract: Commuting plays a crucial role in addressing global climate challenges. This study aims to investigate the influence of households on commuting behaviors. By integrating common household characteristics and employing the TabNet model to analyze daily travel data of 5663 residents in Wenling City, it was found that household characteristics play a more significant role in commuting carbon emissions for most travelers compared to individual characteristics. Specifically, the primary mode of transportation within the household, the ownership of transportation vehicles, and the average number of trips per household member predominantly influence this impact. The results also indicate that this influence is less pronounced among socially vulnerable groups (women, minors, elderly, and low-income individuals), while varying degrees of similarity are observed among men, middle-aged individuals, and high-income groups. Based on these findings, this study suggests that the formulation of low-carbon policies should be tailored to the household circumstances of different population groups, and the effectiveness of low-carbon emission policies can be evaluated based on changes in household situations.