|
Activity Number:
|
376
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 4, 2009 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Business and Economic Statistics Section
|
| Abstract - #304679 |
|
Title:
|
Multifrequency Forecasting with SAS High-Performance Forecasting Software
|
|
Author(s):
|
Michele A. Trovero*+ and Ed Blair and Micheal J. Leonard
|
|
Companies:
|
SAS Institute Inc. and SAS Institute Inc. and SAS Institute Inc.
|
|
Address:
|
1430 HWY 206, Bedminster, NJ, 07921,
|
|
Keywords:
|
forecasting ; benchmarking ; multifrequency ; accumulation
|
|
Abstract:
|
Forecasters often deal with data accumulated at different time intervals (for example, monthly data and daily data). A common practice is to generate the forecasts at the two time intervals independently so as to choose the best model for each series. That practice can result in forecasts that do not agree. The forecasts generated with the less frequent data are usually more reliable in the long term because the number of multistep predictions in the forecasting horizon is smaller. Additionally, the higher-frequency data might display intermittent features that make the application of traditional time series models impractical. This paper shows how the SAS High-Performance Forecasting HPFTEMPRECON procedure uses the low-frequency forecasts as a benchmark to adjust the higher-frequency forecast to take best advantage of both forecasts.
|
