444 / 2019-02-27 20:58:52

Estimation of Pedestrian Density and Speed on Street Network Using Smartphone Spatio-Temporal Data

Mobile Phone,GPS,Human Mobility,Street Pattern,Space Syntax

全体主题 > Big data in Urban Planning and Management

Today, people movement data collected from mobile phones are used for providing various services or for marketing. However, there are not many studies investigating pedestrian density and speed using such data. In this study, we aim to quantitatively describe the relation of street network configuration, potentials from a railway station, land use, street form such as width or length, to the density and the speed of pedestrians using correlation analysis and multiple regression analysis. The results of these analyses provide the model of how we walk in urban space and can be applied to urban planning, urban development or store opening planning.
In this research, we used XY point data from smart phone1, pedestrian traffic data2, the number of train station user data3 and digital map of road center line, railway center line and land use4. The point data of people flow were collected from smartphone passing through Saitama City throughout a year. Concretely, the data set contains approximately 5,000 anonymous users and 500,000 points per day. Each record in the dataset has information of daily user ID, latitude, longitude, time-stamp, locational accuracy, speed, direction of movement, etc. In this research, we extracted the pedestrian data which were positioned in Omiya, with the accuracy of less than 30 meters’ error, obtained from iOS smartphones. The point data were aggregated at road center line segments separately using GIS, and the number of associated points, the average and standard deviation of movement speed were given to each segment. Then, we conducted space syntax analysis using depthmapX, and the street information was given to each segment.
Finally, we investigated the relationship between characteristics of pedestrian and those of street by correlation analysis and multiple regression analysis. In the analysis of pedestrian density, we examined not only the original pedestrian density but also the logarithmic value of pedestrian density in two sizes of the areas. The large area is as large as Omiya ward, and the small area is limited to within the central districts of Omiya with many commercial facilities. In the analysis of walking speed, we examined average and standard deviation of walking speed.
In the analysis of pedestrian density, one of the best multiple regression models we made is shown in figure 1, where the adjusted R2 of the model was 0.698. We found that the most influential variable is the potential of metric distance to the negative 2nd power and the following variables are the square root of the area of the adjacent commercial or business land-use, the logarithmic Angular Choice of large angular radius We also found that the indices of space syntax are higher corresponding to the average and standard deviation of walking speed and the segments near Omiya station have low average and low standard deviation. We showed one of the results of which the adjusted R2 of the model was 0.625, in figure 2.