Factor Analysis-Based Method for Clustering Taxi Zones.

Clustering taxi zones, a relatively novel approach to urban transportation, has emerged as a viable solution to enhance the efficiency and sustainability of taxi services. This study suggests using a modified factor analysis version augmented with a simple integer programming model (FA-IP) for categorizing taxi zones into clusters. The practicality of this method was demonstrated using a real dataset containing 375,639 taxi trips conducted during the first week of January 2022 across 66 zones in the Manhattan area. This study represents the inaugural application of an FA-IP method for grouping taxi zones. Alongside its anticipated superior performance, this method eliminates the need for randomly selecting initial cluster centroids, distinguishing it from the K-means clustering algorithm, a commonly used clustering algorithm. Another advantage of the method proposed is its stability in contrast to commonly utilized clustering algorithms, which may yield different clusters with each iteration.