10-10-2013 11:11 AM
I know my question might be very general but would there be some kind of guide of what model to try to fit in every case? are there any indication on how to move accordingly?
Thanks in advance
10-10-2013 11:19 AM
This is a very general question indeed - but I'll take a shot at it, maybe you could provide some background for your question.
For forecasting modeling challenges exponential smoothing models will do a good job typically. Of course I'm talking about ESM with optimized parameters, which are available both in SAS/ETS (in PROC ESM) or in SAS Forecast Server. Of course they can fall short as well, in particular if there is too little history available or if your time series patterns are sporadic or intermittent.
Is this helpful at all?
10-10-2013 11:23 AM
Assuming you what kind of variables you have i.e (independent but also categorical, interval..etc) I refer people to UCLA's site they have a pretty good chart for determining what kind of statistical test to use: Choosing the Correct Statistical Test in SAS, Stata and SPSS
10-11-2013 08:19 AM
"Causal time series models are used to forecast time series data that is influenced by causal factors. Input variables (regressor or predictor variables) and calendar events (indicator, dummy or intervention variables) are examples of causal factors. These independent (exogenous) time series causally influence the dependent (response, endogenous) time series, and therefore can aid the forecasting of the dependent time series. Examples of causal time series models are autoregressive integrated moving averages with exogenous inputs (ARIMAX), which are also known as transfer function models, and dynamic regression models and unobserved component models (UCM), which are also known as state-space models and structural time series models."
SAS/ETS will give you access to these models in the following procedures: AUTOREG, ARIMA, UCM, or SSM.
Currently the ESM procedure does not allow for independent variables.