N days average volume based ARIMA forecasting model for Shanghai metro passenger flow

The paper introduced an index of `n-day' moving average passenger flow volume aimed to reflect a more representative and stable daily passenger flow, based on which an ARIMA model is constructed for forecasting daily passenger flow of Shanghai metro. Using a `7-day' moving average volume against actual daily volume, the model calculates 2 new sequences of daily volume separately by iterative and recursive algorithm. The paper analyzed changes of Shanghai metro passenger flow, and the change rate of daily volume against `7-day' average was used for analyzing the sudden changes of passenger flow before and during main holidays. Empirical tests show that the relative error of recursive forecasting is less than that of iterative, and both with a relative error around 2% for forecasting massive passenger flow before and during main holidays.