An efficient data-driven method to construct dynamic service areas from large-scale taxi location data

Abstract

Service area analysis is crucial for determining the accessibility of public facilities in smart cities. However, the acquisition of service areas using conventional approaches has been limited. First, investigating traffic flow is difficult, as this factor varies significantly over time and space. Second, obtaining service areas of mobile facilities/targets has remained a challenge owing to a lack of data and methods. To address these problems, this study proposes an efficient big-data-driven approach that utilizes large-scale taxi GPS location data collected over two years within Seoul City and distributed computation to obtain the average travel time values on fine-grained grid cells of 100 m × 100 m resolution. On-the-fly visualization methods were then established with an ability to construct isochrone maps of service areas in near-real-time. This enabled performing accurate service area analysis of mobile facilities/targets dynamically. The proposed solution can be effectively used in various applications, such as optimizing the ride-sharing services or the routes of autonomous electric vehicles in future smart cities, as demonstrated in this study.