Hourly Pedestrian Population Trends Estimation using Location Data from Smartphones Dealing with Temporal and Spatial Sparsity
22nd ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems, 2014/11
機械学習 (Machine Learning) データサイエンス (Data Science)
- "This paper describes a pedestrian population trend estimation method using location data of smartphone users. This technique is intended to be an alternative to traffic censuses using tally counters. Traffic censuses using tally counters are still commonly used to survey the number of pedestrians despite their cost and limitations in area and time. The proposed approach can replace the traffic census by using smartphone users' location data accumulated on Yahoo! Japan. Moreover, it is low cost because it uses location data collaterally acquired from smartphone users, and it has no limits in terms of area or time. This means pedestrian population trends in arbitrary areas and times about which we want to know can be estimated. The proposed technique is based on the assumption that the number of location data in an area is proportional to the population volume, but it also eliminates some data to increase pedestrian accuracy. In the elimination step, some location data that should not be counted as pedestrians are excluded by estimating transport modes from anteroposterior location data. The supplement step tackles the problem of data shortage when a target area is a small region by using a Gaussian kernel. The Gaussian kernel smoother is also used to deal with data interpolation in the time direction, and it enables us to estimate time-continuous pedestrian volumes in arbitrary areas. To evaluate the approach, a manual traffic survey was conducted in five areas on 11 days and the ground truth data are acquired. Experimental result shows the approach successfully estimate pedestrian population trends in areas. The proposed method makes less than one-tenth the mean squared errors of hourly pedestrian number estimation than the conventional approach."
Hourly Pedestrian Population Trends Estimation using Location Data from Smartphones Dealing with Temporal and Spatial Sparsity（外部サイト／External Site Link）