| 摘要 |
<p>A method and computer systtem is provided for automatically constructing a time series model for the time series (figure 2). The model can be either a univariate or multivariate ARIMA model, depending upon whether predictors, interventions or events are input in the system in addition to the times series. The method for constructing the univariate ARIMA model comprises the steps of inputting missing values of the corresponding times series, finding the proper transformation for positive time series, determining differencing orders, determining non-seasonal AR and MA oders by pattern detection, building an ininitial model, and iteratively estimating and modifying the model. The method for constructing the multivariate model comprises the steps of finding a univariate ARIMA model for the time series, applying the transformation found in the univariate model to all positive time series including the series to be forecast and predictors; applying differencing orders found in the univariate model to all time series including the series to be forecast, predictors, interventions and events, deleting selected predictors and further differencing other predictors, building an initial model wherein its disturbance series follows an ARIMA model with AR and MA orders found in the univariate model, eand iteratively estimating and modifying the model.</p> |
