大型活动事件分级评价数据集

为评价大型活动扰动事件对周边地面交通系统带来的影响，选取了福田区会展中心周边的福华路、福华一路、金田路、益田路作为数据采集点，这些路段在深圳会展中心举办活动的日期内覆盖率最高。在预处理阶段，拟采用多种方式进行数据的质量校验：首先对异常值进行筛选和校验，将明显偏离正常值的记录删除。同时，针对原始道路数据集中存在的缺失和空值问题，通过基于时间序列的插值以及历史数据的均值估计填补缺失的数据。通过查找公开官方数据，锁定会展中心举办大型活动的日期。针对这些日期，根据填补后四条道路的流量数据表，分别计算以下两个指标：(1)路网平均等待时间：举办活动期间特定时段，道路的平均等待时间，即所选路网全天的车辆平均等待时间与道路数的比值；（2）道路总流量变化率：举办活动期间道路总流量和无活动时候相比的变化率。此外，在进行指标计算时，考虑到周内各天之间的交通需求差异，对工作日和非工作日分别进行计算，以保证数据的有效性。将以上两个指标输入所构建的基于改进密度峰值聚类的扰动事件分级评价模型，对大型活动造成的影响进行评估，得到最终的分级评价结果。

To evaluate the impact of disturbance events induced by large-scale activities on the surrounding ground traffic system, Fuhua Road, Fuhua First Road, Jintian Road and Yitian Road around the Convention and Exhibition Center in Futian District were selected as data collection points. These road sections have the highest data coverage during the dates when events are held at the Shenzhen Convention and Exhibition Center.

In the preprocessing stage, multiple methods are employed for data quality validation: first, outliers are screened and verified, and records that significantly deviate from normal values are deleted. Meanwhile, to address the missing and null values in the original road traffic dataset, missing data is filled via time series interpolation and mean estimation based on historical data. Official public data is searched to identify the dates when large-scale events are hosted at the Convention and Exhibition Center.

For these dates, the following two indicators are calculated separately based on the imputed traffic volume datasets of the four roads: (1) Average Road Network Waiting Time: The average waiting time of roads during specific time windows of the event period, specifically the ratio of the average vehicle waiting time of the selected road network over the entire day to the number of roads included in the study; (2) Change Rate of Total Road Traffic Volume: The relative change rate of the total road traffic volume during the event period compared to that during non-event periods.

In addition, given the differences in traffic demand across different days of the week, calculations are conducted separately for weekdays and non-weekdays to ensure data validity. The two aforementioned indicators are fed into the developed improved density peak clustering-based graded evaluation model for disturbance events to assess the impact of large-scale activities, yielding the final graded evaluation results.