Analysis of the passenger pick-up pattern for taxi location recommendation

This paper analyzes a pick-up pattern of taxi service in Jeju area based on the real-life location history data collected from the Taxi Telematics system, aiming at obtaining useful background data necessary to design a location recommendation service for empty taxis. Out of the great amount of location records, pick-up data are extracted by tracing the state change in the predefined taxi state diagram. To decide a reasonable granularity of location recommendation, refined clustering is performed by means of the well-known k-means method supported by the E-Miner statistics software package. In addition, within each cluster, the temporal analysis creates time-dependent pick-up pattern change along the time axis. As a result, the cluster and its spatio-temporal pick-up frequency make it possible to suggest that the empty taxi go to the nearby cluster location, resulting in the reduction of empty taxi ratio.