This study examines the school travel mode of children and youth students (ages 7 to 18) in Singapore. Using a large crowdsensing dataset the paper focuses on minute-by-minute decision-making of those students living within $2.5$km of school. Data-driven methods are employed in order to identify students’ chosen transport mode (car, walking, taking bus or riding metro). Furthermore, we present attributes of travel mode alternatives computed by a replicable framework that utilises open sources. New algorithms are developed to identify proxies for walking access and public transport access. We found that about 19% of students in the sample live up to a distance of $2.5$km from the school. From these, about 45% of trips are made by public transit (e.g., bus and metro), and only 13% are made by walking. The empirical results suggest that the public transport modes of bus and metro are not distinct. Consistent with past research based on traditional survey data, walking time and walking distance are the most influential factors in the decision to walk-to-school. Interestingly, schools’ connectivity to the street network is found to play a key role on the shift from public transport to walking. Likewise, departing at peak hours, the odds to choose public transport modes are about 40-45% lower as compared to walk.

本研究调查了新加坡儿童和青年学生（7 至 18 岁）的学校出行方式。该论文使用大型人群感知数据集，重点关注生活在其中的学生的每分钟决策$2.5$公里的学校。采用数据驱动的方法来识别学生选择的交通方式（汽车、步行、乘坐公共汽车或乘坐地铁）。此外，我们介绍了由利用开源的可复制框架计算的出行方式替代方案的属性。开发了新的算法来识别步行访问和公共交通访问的代理。我们发现样本中约有 19% 的学生生活距离为$2.5$离学校公里。其中，约 45% 的出行是通过公共交通（例如，公共汽车和地铁）进行的，只有 13% 是通过步行进行的。实证结果表明，公共汽车和地铁的公共交通方式并不明显。与过去基于传统调查数据的研究一致，步行时间和步行距离是决定步行上学的最重要因素。有趣的是，我们发现学校与街道网络的连通性在从公共交通到步行的转变中发挥了关键作用。同样，在高峰时段出发，选择公共交通方式的几率比步行低约 40-45%。