Introduction

Time Series Forecasting System

SAS/ETS software includes the Time Series Forecasting System , a point-and-click application for exploring and analyzing univariate time series data. You can use the automatic model selection facility to select the best-fitting model for each time series, or you can use the system’s diagnostic features and time series modeling tools interactively to develop forecasting models customized to best predict your time series. The system provides both graphical and statistical features to help you choose the best forecasting method for each series.

The system can be invoked by selecting Analysis→Solutions, by the FORECAST command, and by clicking the Forecasting icon in the Data Analysis folder of the SAS Desktop.

The following is a brief summary of the features of the Time Series Forecasting system. With the system you can:

  • use a wide variety of forecasting methods, including several kinds of exponential smoothing models, Winters method, and ARIMA (Box-Jenkins) models. You can also produce forecasts by combining the forecasts from several models.

  • use predictor variables in forecasting models. Forecasting models can include time trend curves, regressors, intervention effects (dummy variables), adjustments you specify, and dynamic regression (transfer function) models.

  • view plots of the data, predicted versus actual values, prediction errors, and forecasts with confidence limits. You can plot changes or transformations of series, zoom in on parts of the graphs, or plot autocorrelations.

  • use hold-out samples to select the best forecasting method

  • compare goodness-of-fit measures for any two forecasting models side-by-side or list all models sorted by a particular fit statistic

  • view the predictions and errors for each model in a spreadsheet or view and compare the forecasts from any two models in a spreadsheet

  • examine the fitted parameters of each forecasting model and their statistical significance

  • control the automatic model selection process: the set of forecasting models considered, the goodness-of-fit measure used to select the best model, and the time period used to fit and evaluate models

  • customize the system by adding forecasting models for the automatic model selection process and for point-and-click manual selection

  • save your work in a project catalog

  • print an audit trail of the forecasting process

  • save and print system output including spreadsheets and graphs