On discovering regional taxi service disequilibrium with geographical collaborative filtering

A challenging problem that taxi service faces is to fulfill all passenger requests in different regions of a city and different time periods. Taxicab service rate of a region calculated from existing trajectory data can indicate the utilization rate of taxicabs in the region and help solving the problem. However, as trajectory data often contain corrupt or missing records in real scenarios, taxicab service rates of some region may not be available. This paper proposed a method to estimate taxicab service rates based on collaborative filtering with geographical factors and improved similarity measurements. The method proves to be efficient after validation on a large taxicab trajectory dataset.