Taxi origin-destination areas of interest discovering based on functional region division

This paper introduced a new method to discover the taxi Origin-Destination areas of interest based on functional region division to resolve the problem that taxis distribution is unreasonable. The problem occurs due to dispatching taxis is not in line with the taxi demand in different functional regions during different time periods. The method proposed in this paper clustered taxi Origin-Destination data points of the functional region with the improved DB-SCAN algorithm and yielded the areas of interest with the statistical method named Chi-square distribution. The accuracy and credibility of the method was evaluated through two indicators called coverage and hit ratio and a comparison analysis on situations of different regions. The experiment proved that the proposed method can effectively discover and predicted the taxi Origin-Destination areas of interest in different functional regions in different time periods.