With the development of technology and urbanization, cars have greatly facilitated people's travel, but they also bring problems such as road congestion, pollution and accidents, which restrict the healthy development of cities. How to predict traffic flow efficiently and accurately has become a hot topic. Taxi data is highly recommended in urban transportation research due to the characteristics of broad spatial coverage, long driving time and relatively complete data. In this paper, the data of yellow taxis in New York during March 2023 are used to present the demand changes by means of line chart thermal map, and divide the database of March into two parts, one is training set and the other one is test set, then simulation and analysis using ARIMA model. It reveals the changing relationship between taxi demand and time. By predicting the traffic flow, this study can alleviate the congestion of urban main roads and improve the quality of citizens’ life. At the same time, forecasting commuting demand can balance supply and demand, provide more efficient services for operating companies, taxi drivers and passengers, and provide decision-making basis for government planning.