Instituto de Física, Universidad Nacional Autónoma de México, Ciudad Universitaria, 04510, Mexico City, Mexico.
Sci Rep. 2022 Jan 7;12(1):98. doi: 10.1038/s41598-021-04037-6.
In this paper, we analyze a massive dataset with registers of the movement of vehicles in the bus rapid transit system Metrobús in Mexico City from February 2020 to April 2021. With these records and a division of the system into 214 geographical regions (segments), we characterize the vehicles' activity through the statistical analysis of speeds in each zone. We use the Kullback-Leibler distance to compare the movement of vehicles in each segment and its evolution. The results for the dynamics in different zones are represented as a network where nodes define segments of the system Metrobús and edges describe similarity in the activity of vehicles. Community detection algorithms in this network allow the identification of patterns considering different levels of similarity in the distribution of speeds providing a framework for unsupervised classification of the movement of vehicles. The methods developed in this research are general and can be implemented to describe the activity of different transportation systems with detailed records of the movement of users or vehicles.
在本文中，我们分析了 2020 年 2 月至 2021 年 4 月墨西哥城快速公交系统 Metrobús 中车辆行驶的大量数据集。通过这些记录以及系统划分为 214 个地理区域（路段），我们通过对每个区域的速度进行统计分析来描述车辆的活动。我们使用 Kullback-Leibler 距离来比较每个路段的车辆运动及其演变。不同区域的动力学结果表示为一个网络，其中节点定义系统 Metrobús 的路段，边缘描述车辆活动的相似性。该网络中的社区检测算法允许在考虑速度分布相似性不同水平的情况下识别模式，为车辆运动的无监督分类提供了一个框架。本研究中开发的方法是通用的，可以用于描述具有用户或车辆行驶详细记录的不同交通系统的活动。
