ESTIMATING SPATIOTEMPORAL DISTRIBUTION OF MOVING PEOPLE BY INTEGRATING MULTIPLE POPULATION STATISTICS

DOI Web Site Web Site 1 Citations 14 References Open Access

Bibliographic Information

Other Title
  • 携帯電話人口統計を用いた都市内移動者の時空間分布推定
  • ケイタイ デンワ ジンコウ トウケイ オ モチイタ トシ ナイ イドウシャ ノ ジクウカン ブンプ スイテイ

Search this article

Abstract

<p> There is a growing demand for data that allow a highly accurate understanding of the spatiotemporal distributions of both moving and static people in urban areas. Currently, a variety of population data are available, but none of such data provide an accurate understanding of numbers and directions of moving people based on detailed units of space and time. In this paper, by integrating multiple sets of data, including Mobile Spatial Statistics (MSS), Congestion Statistics (CS), and Person Trip survey data (PT data), we constructed a method of estimating the number of people who flow in and flow out separately in detailed units of space and time, by considering the advantages and disadvantages of each type of population statistics. Furthermore, we demonstrated the characteristics of the spatiotemporal distribution of moving people vary according to regional characteristics of areas, day of week, and time.</p><p> The number of people in each grid-cell at a unit time, which is obtained every 60 minutes from MSS, does not distinguish between moving (inflow or outflow) and static occupants. Therefore, the ratio of static people to total people (static people ratio) in each grid-cell, at each time, was estimated using the number of static people and the total number of people obtained from PT data at 60-minute intervals, by assuming the number of static people is proportional to the total floor area of each building by building use. Next, using the estimated static-people ratio, the number of moving people and static people were estimated by a maximum likelihood method with constrains, in which the total estimated number of people (inflows, outflows, and static occupants) always maintain consistency with the number people obtained from MSS.</p><p> In the CS, the number of people moving between grid-cells in a 5-minute period is obtained as an aggregated value every 60 minutes. Based on this dataset, the movement probability between grid-cells in a 5-minute period was estimated and the movement probability matrix was constructed. The number of people moving between grid-cells was estimated by a maximum likelihood method using the movement probability. In this process, the number of inflows and outflows people estimated above were used as constrains.</p><p> Using the estimated database on inflow/outflow/static people, we discussed the spatiotemporal characteristics of the number, direction, and distribution of moving people, which varies according to areas, day of week, and time. More specifically, we demonstrated that (a) in central business districts, there are more inflows than outflows in the morning, that both flows are about the same during the daytime, and that the outflow increases in the evening; and (b) in high-density residential districts and areas including large universities, we found that there are more outflows than inflows in the morning, and that both flows are almost the same in the daytime and evening.</p><p> Finally, we estimated the spatiotemporal distribution of inflows/outflows of people on bank holidays. Within the 23 wards of Tokyo, we found that more people move during the day time on bank holidays than on weekdays and differences in movement direction in central business districts were demonstrated.</p><p> In conclusion, we demonstrated that the proposed method of analyzing the characteristics of moving people on a micro-spatiotemporal scale has significant potential for use in a variety of fields that include transportation planning, environmental planning, and disaster mitigation planning, as well as geo-marketing.</p>

Journal

Citations (1)*help

See more

References(14)*help

See more

Related Projects

See more

Keywords

Details 詳細情報について

Report a problem

Back to top