11-29-2014 11:02 AM
I am trying to leverage best available option to forecast time series data using SAS procedures.
I am trying and testing Proc Forecast and Proc ESM from last few days and I came across many differences in capabilities, usability, limitations, end output production and underlying algorithms or logic for selecting automatic model when no any model is specified.
Primarily I am seeking to use any of these procedures such that it will automatically select the best fit model for all of the variables in the dataset by analyzing the pattern of data in the past for individual variable, can anyone please enlighten scope of using these procedures for the purpose I stated.
Also what are pros and cons of using any of the Procedure.
12-18-2014 02:13 PM
per this site - proc forecast is more or less deprecated so I would not write any new code that depends on it. They recommend using ESM or autoreg instead.
"The FORECAST procedure is superseded by newer SAS/ETS procedures that provide more powerful and flexible versions of the forecasting methods provided by PROC FORECAST"