Forecasting Passenger Flow in Chengdu Metro Based on the ARIMA Model

With the increase in population, short-term forecasting of subway traffic is becoming increasingly important. Based on data from the first half of 2025, Chengdu metro ridership, the system is constructed by using the autoregressive integrated moving average (ARIMA) model. The study first confirms the statistical properties of the passenger flow series through the Augmented Dickey-Fuller (ADF) smoothness test, and then uses autocorrelation and partial autocorrelation analyses to determine the model order, establishes the ARIMA (1,1,1)(1,0,1) seasonal prediction model, and the prediction model has a good mean square error (MSE) value of 1910.41. The prediction shows that the passenger flow will maintain the growth trend and keep the rules fluctuating. The results provide a basis for the operations department to rationalize vehicle schedules.