Uncertain Autoregressive Moving Average Analysis with Application to Metro Passenger Flow Forecast

When the residuals sequence generated by the traditional autoregressive moving average (ARMA) model is not a random sequence, the uncertain hypothesis testing method is used to test the residuals sequence. If the residuals sequence passes the uncertain hypothesis testing, it can be regarded as an uncertain sequence composed of uncertain variables. Then uncertain autoregressive moving average (UARMA) model can be used to fit the residuals sequence. After that, based on the uncertain statistics methods, the corresponding point prediction and interval prediction can be performed, which can provide more favorable support for the decision makers. Finally, the real passenger flow data of London underground are analyzed and predicted, and the results illustrate that the UARMA model is more reasonable and effective than traditional ARMA model.