As an important mode of travel for urban residents, taxis play an important supplementary role to the public transportation system, and it can cooperate with ground public transportation, rail transit and other transportation methods to make residents' travel more efficient and convenient. The taxi equipped with GPS system can collect all-day driving data and operation data, which can clearly reflect the time-varying characteristics of urban residents' travel. However, due to the huge amount of GPS data in the entire city's taxis throughout the day and the complex data structure, it is difficult to directly obtain residents' travel rules from the massive data. Based on the GPS data of a certain day taxi in Shanghai, this paper studies the massive data processing and mining method on the basis of analyzing the data structure, and maps the data to the road network of the whole city to analyze the temporal and spatial distribution rule of urban taxis, and further studies the influencing factors of the time variation of taxi passengers getting on and off the taxi, so as to provide theoretical support for the reasonable scheduling and operation of taxis. Keywords: Taxi; Passengers; Time and space distribution; GPS trajectory; Big data

CICTP