Pseudo-PFLOW: Development of nationwide synthetic open dataset for people movement based on limited travel survey and open statistical data
Pseudo-PFLOW: Development of nationwide synthetic open dataset for people movement based on limited travel survey and open statistical data
复制标题
DOI：
10.48550/arxiv.2205.00657
复制
发表时间：
2022-05
期刊：
影响因子：
--
通讯作者：
Takehiro Kashiyama;Y. Pang;Y. Sekimoto;T. Yabe
中科院分区：
文献类型：
--
作者：
Takehiro Kashiyama;Y. Pang;Y. Sekimoto;T. Yabe
人们流数据用于城市和商业计划和灾难管理等各个领域。但是，由于隐私问题，很难获得从移动电话收集的人流量数据，例如使用全局定位系统和呼叫详细记录数据。即使获得了数据，它们也很难处理。这项研究通过结合了有限城市地区的公共统计和旅行调查数据，开发了伪流数据，涵盖了日本的整个日本。该数据集不是实际旅行运动的代表，而是人们的典型工作日运动。因此，预计它将用于各种目的。此外，与旅行调查不同，数据集代表了整个日本的无缝运动，对覆盖范围没有任何限制。在本文中，我们提出了一种生成伪people-flow的方法，并描述了覆盖大约1.3亿人口的“伪级”数据集的开发。然后，我们使用来自多个大都市地区的手机和旅行调查数据评估了数据集的准确性。结果表明，确认了超过0.5的确定系数，以进行有关种群分布和跳闸量的比较。晚上8:00对于每个量表图10。为了确认行为模型的有效性，R2值在每个时间段的手机数据和伪people-flow数据显示在上午6:00时显示。结果显示，该结果没有显示出准确性的变化。一天中的一天中的时间。比管理边界级别小的评估表明，凌晨6:00，手机数据和伪流 - 流数据之间的比较结果在许多人处于活动状态时更糟。结果表明，行为模型复制了人的运动，捕获了人口分布的变化。
People flow data are utilized in diverse fields such as urban and commercial planning and disaster management. However, people flow data collected from mobile phones, such as using global positioning system and call detail records data, are difficult to obtain because of privacy issues. Even if the data were obtained, they would be difficult to handle. This study developed pseudo-people-flow data covering all of Japan by combining public statistical and travel survey data from limited urban areas. This dataset is not a representation of actual travel movements but of typical weekday movements of people. Therefore it is expected to be useful for various purposes. Additionally, the dataset represents the seamless movement of people throughout Japan, with no restrictions on coverage, unlike the travel surveys. In this paper, we propose a method for generating pseudo-people-flow and describe the development of a "Pseudo-PFLOW" dataset covering the entire population of approximately 130 million people. We then evaluated the accuracy of the dataset using mobile phone and trip survey data from multiple metropolitan areas. The results showed that a coefficient of determination of more than 0.5 was confirmed for comparisons regarding population distribution and trip volume. 8:00 p.m. for each of scales Figure 10. To confirm the effectiveness of the behavioral model, the R2 values are shown with mobile phone data for each time period and pseudo-people-flow data at 6:00 a.m. The results show no change in accuracy by the time of day at either scale. The evaluation at scales smaller than the administrative boundary level shows that the comparison results between the mobile phone data and pseudo-people-flow data at 6:00 a.m. is worse toward midday when many people are active. The results indicate that the behavioral model replicates the movement of people, capturing changes in population distribution.