学区尺度下小学生通学事故风险评估及影响因素

为评估学区尺度下小学生通学事故风险并获取其影响因素，融合交通事故数据、道路运行数据和学区划分数据，构建学区尺度下通学事故风险评估方法，并运用随机森林模型分析其影响因素。以小学生在学区内部小学通学为原则，构建基于交通事故数据和道路长度的道路暴露度模型评估小学生通学事故风险。以深圳市中心城区为例进行验证。结果表明：深圳市中心城区高风险学区主要聚集于南山区北部和罗湖区北部，低风险学区主要分布在南山区中南部和福田中部学区；基于随机森林构建的学区尺度下小学生通学事故风险评估模型预测准确率达85.93%，能够较为准确地评估小学生通学事故风险；小学密度和学区面积是小学生通学事故风险的主要影响因素，分别能解释37.15%和22.86%通学事故风险。

Risk Assessment and Influencing Factors of Pupils' School Commuting Accident Risk in School District Scale

This paper proposes a risk assessment method of school commuting accident and identify the crucial influencing factors, by integrating traffic accident data, road operation data and school district zoning data, the study developed the accident assessment method and analyzed the influencing factors using the random forest model. With the assumption that pupils' school commuting is within the school district, a road exposure model was developed through traffic accident data and road length data to evaluate the pupils' school commuting accident risk. A verification analysis was conducted by taking the central city of Shenzhen as an example. The results show that: the high- risk school districts in the central city of Shenzhen are mainly in the north of Nanshan and Luohu, while the low-risk school districts are mainly distributed in the south-central part of Nanshan and the central part of Futian. The risk assessment model based on the random forest produced 85.93% prediction accuracy. Which shows the performance of the proposed model for the pupils' school commuting accident risk assessment. Primary school density and school district zoning area are two major influencing factors of pupils' school commuting accident risk, which can explain respectively 37.15% and 22.86% of school commuting accident risks.