Taxi Hotspots Identification through Origin and Destination Analysis of Taxi Trips using K-means Clustering and H-indexing

Abstract

It is apparent that the taxi industry has grown and developed over the years. In addition to that, it will presumably continue to grow as time goes on due to the increasing popularity of taxi-hailing applications. However, taxi origin and destination (O-D) locations are not clearly established since taxis are very flexible in terms of where they can pick up and drop off passengers. In this study, the taxi origin and destination hotspots are determined by first clustering the available O-D pairs from empirical mobility traces. The validity of these formed clusters is determined by utilizing the silhouette analysis. Finally, hotspots are located by measuring the cluster’s h-index. Simulation results reveal that more clusters tend to provide unreliable silhouette values due to the fact that origin/destination GPS points are very close to each other. For a given number of clusters, the h-index tend to locate clusters that can be considered as hotspots.

Hotspot Identification Through Pick-Up and Drop-Off Analysis of Ride-Hailing Transport Service

“…By identifying origin and destination (OD) flow clusters in urban travel data, it is possible to determine prospective routes for public transportation service settings [8]. In order to locate the taxi OD hotspot, the available OD pairs from empirical mobility traces are first grouped [9]. A deep understanding of hotspots, namely areas with heavy travel activity, has great potential for shaping more efficient transportation planning and better traffic management [10].…”

Mining passenger-seeking road patterns of high-occupancy taxis based on cross-tabulation analysis

“…Gao Zhan [1] et al calculate the core points of taxi loads by circular slicing the Origin-Destination data of taxi orders, combining with the area of the region and the number of orders to model, and propose a gridded Manhattan path algorithm for empty taxis to search for passengers, gridded the region between taxis and the core points of loads, and find out the Manhattan path with the largest probability of loads to be recommended to the empty taxi drivers; FENG Hui-fang [2] and others proposed similarity metric algorithm to calculate the spatial similarity, temporal similarity and spatial similarity of the core trajectories, and combined with DBSCAN clustering algorithm to cluster the passenger trajectories, and according to the clustering results to obtain the spatial distribution of the urban hotspot passenger paths; Magsino [3] et al identified taxi starting and ending hotspots by clustering available O-D pairs from empirical mobility trajectories, then used contour analysis to determine the effectiveness of these formed clusters, and finally located the passenger-carrying hotspot areas by measuring the h-index of the clusters. However, few domestic and international studies have explored the taxi high occupancy path scheme from the perspective of the relationship between taxi occupancy and hotspot attraction, road class attraction, and other factors.…”