【摘要】 为响应国家绿色出行的号召，依据城市居民绿色出行意愿来探究城市居民绿色出行行为的关键影响因素。通过问卷调查获取影响城市居民绿色出行行为的潜在因素，采用过滤式filter和包装式wrapper两种特征选择算法对关键因素进行初步提取：结合filter算法中的信息增益率与基尼指数增益作为评估特征变量的标准对各因素进行评分排序，在wrapper算法中构建多项式朴素贝叶斯、随机森林以及支持向量机3种分类器模型，选择分类性能较好的分类器筛选特征变量。利用多元Logistic回归模型对筛选出来的因素进行2次提取，参考显著性水平以及估算系数分析各因素对绿色出行行为的具体影响。结果表明：群体出行的出行方式、出行态度、疫情影响、出行费用、环保意识、主观规范、学历以及私家车拥有情况8个因素对城市居民绿色出行行为具有较强的影响，并提出针对性的建议，为交通管理者倡导城市居民绿色出行行为提供参考。更多还原

【Abstract】 In response to the national call for green travel, this paper explores the key influencing factors of green travel behavior according to the green travel willingness of urban residents. The potential factors affecting the green travel behavior of urban residents were obtained through questionnaire survey. The key factors are preliminarily extracted using filter and wrapper feature selection algorithms. In filter feature selection algorithms, a combination of the information gain rate and Gini index gain are used as the criteria to evaluate the characteristic variables. Three classifiers, MNB, RF and SVM, are constructed, and the classifiers with better classification performance are used to screen the characteristic variables in wrapper feature selection algorithms. Multiple Logistic regression model is used for secondary extraction of the screened factors, the specific impact of each factor on green travel behavior is analyzed by reference to the significance levels and the estimated coefficients. The results showed that eight factors, such as the companion travel methods, travel attitude, impact of the epidemic, travel cost, environmental awareness, subjective norms, educational background and private car ownership, had a strong impact on green travel behavior of urban residents, and put forward the specific suggestions to provide a reference for traffic managers to advocate urban residents to select green travel.更多还原