SAS Support Communities

SnurreJensen · ‎08-31-2015

I would recommend starting with plotting your timeseries - both the original series, but also the (seasonally) differenced series and the seasonal cycles. This will give you an indication of the patterns in your data. Plotting can be done using PROC TIMESERIES which is also part of SAS/ETS. For more info, please have a look at the documentation: The TIMESERIES Procedure :: SAS/ETS(R) 14.1 User's Guide Once that is done, start using the identify statement in PROC ARIMA to examine which orders of differencing you need - if any, type of lag-structure etc. Once you have that in place you can start producing forecasts. For more information on PROC ARIMA please consult the documentation which can be found here: The TIMESERIES Procedure :: SAS/ETS(R) 14.1 User's Guide Thanks, Snurre

SnurreJensen · ‎08-28-2015

You are correct - the use of the FORECAST statement in PROC IMSTAT requires a separate license. Thanks, Snurre

SnurreJensen · ‎08-13-2015

Do you know this EM tip - https://communities.sas.com/docs/DOC-10125 ? Near the end you there is a discussion on the exp smoothing node which also shows some of the options. Here you need to change the "interval" option.

SnurreJensen · ‎08-12-2015

Which version are you on? You should be able to change that in the Time Series Exponential Smoothing Train Properties. There is an option to specify the time series interval.

SnurreJensen · ‎08-12-2015

Hi, I'm a bit confused to your reference to a "node". Which application are you using? SAS Enterprise Miner? Usually the default weekend is Saturday and Sunday. However, this can be changed if you use the weekday67w interval. The 6 is the Friday and the 7 is the Saturday. Thanks, Snurre

SnurreJensen · ‎08-10-2015

It is correct that starting with the 7.2 release, SAS Visual Statistics capabilities are fully integrated into SAS Visual Analytics Explorer. The license fee for the additional SAS Visual Statistics capabilities is lower than the license fee for SAS Visual Analytics.

SnurreJensen · ‎08-10-2015

The choice of which ADF test to use depend on the original series as you mention. A series with apparent trend should look to the test including trend etc. This is to ensure that you are testing the null hypothesis against the most relevant alternative. Some more elaborate explanations can be found here: http://stats.stackexchange.com/questions/44647/which-dickey-fuller-test-should-i-apply-to-a-time-series-with-an-underlying-mode

SnurreJensen · ‎08-10-2015

The holdout sample is used exactly as you describe. The out-of-sample calculates fit statistics for a period at the end of the series for just the selected model. In terms of usage I tend to use the holdout sample for large scale automatic model building and the out-of-sample feature when doing manual model building.

SnurreJensen · ‎06-12-2015

Hi, Please bear in mind that when building a good forecast model the aim is not to find a model that fit all the existing data points. The aim is to find a model that is able to provide as accurate forecasts as possible given the exisitng conditions. For instance, if data is very noisy/erratic/stochastic etc. forecast accuracy will be lower than if the has nice stable patterns. Having said that you can work with multiple seasonal cycles in FS by building your own model. For instance an ARIMA model with differences of (7,28) can be used to work with data where there are seasonal patterns each 7 and 28 days (assuming you have obs for all 7 days a week). Thanks, Snurre

SnurreJensen · ‎06-11-2015

By looking at the attached graph my first guess is that you have some periods where the absolute percentage (APE) error for the individual observations are extremely large due to very low (but not zero!) actual values. It's somewhat hard to tell from looking at the graph but what is the forecast and the actual value for 01jan2014 for instance? Try calculating the APE for that period for instance. To counter this, either enable automatic outlier detection (I assume you use SAS Forecast Studio) or create your own model where you include events for those periods where the actuals are extremely low. Thanks, Snurre

SnurreJensen · ‎06-08-2015

Yes. PROC ESM works with one type of exponential smoothing model and optimizes its parameters. In your case you would then need to create multiple runs of PROC ESM for all variables and store the results in order to compare them and select the best model for each series. For your current setup, instead of using the FORECAST command you should probably try the %FORECAST macro instead. More information available here: http://support.sas.com/documentation/cdl/en/etsug/67525/HTML/default/etsug_tfcomref_toc.htm Thanks, Snurre

SnurreJensen · ‎06-08-2015

Hi, I'm not entirely sure what you are using TSFS for but there are a couple of options you could consider: - Use the SAS Macro facility to build some smart code - for instance a loop around the same command where you only substitute the variable name. - Use something like PROC ESM which supports variable lists in the FORECAST statement. Even though the first option is certainly doable, it might be a somewhat daunting task if you are not familiar with the SAS Macro facility already. Unless there are some considerations which your opening post do not mention that should be taken into account, I would recommend the second option. Thanks, Snurre

SnurreJensen · ‎06-01-2015

Not sure if this is the case in this situation but I've experienced similar cases where the root cause was that the Forecast Server license needed to be upgraded. Since the metadata layer is involved, please make sure that any software update is done using the SAS Deployment Wizard. Thanks, Snurre

SnurreJensen · ‎05-29-2015

In cases where storage becomes an issue I would suggest looking into the following possibilities: - Make sure that there are no old projects which are no longer needed that takes up storage space - If the components datasets are not used, avoid creating them at project creation time. In the GUI it is done in Forecasting Settings -> Forecast. If using the %FSCREATE macro this can be controlled using the CREATEOUTCOMPONENT macro variable. Please not that this needs to be defined at project creation time. Additionally, SAS can also work with compression of the data sets that are being used. This is done using the system option COMPRESS. However, this can impact performance since data sets need to be un-compressed as well. Therefore I would recommend that you investigate the 2 first possibilities mentioned before doing anything else. Thanks, Snurre

SnurreJensen · ‎05-26-2015

I have to admit I'm not 100 percent sure what you are asking. A more detailed description of what PROC FORECAST does can be found here: http://support.sas.com/documentation/cdl/en/etsug/67525/HTML/default/etsug_forecast_details03.htm However, if you are looking for a procedure that can optimize the smoothing weights for you based on data I would encourage you to look into PROC ESM which is another procedure in the SAS/ETS package - http://support.sas.com/documentation/cdl/en/etsug/67525/HTML/default/etsug_esm_overview.htm Thanks, Snurre

Online Status	Offline
Date Last Visited	‎09-21-2015 10:55 AM

SAS Support Communities

Re: SEASONAL ARIMA

Re: How to make forecast in report interactive?

Re: About Exponential smoothing time interval

Re: About Exponential smoothing time interval

Re: About Exponential smoothing time interval

Re: SAS VA and SAS Visual Statistics

Re: ADF test in Proc Arima

Re: Out-of-sample range vs holdout sample

Re: Too much of Forecasting Error due to input data

Re: Too much of Forecasting Error due to input data

Re: SAS VA and SAS Visual Statistics

Re: Time Series Forcasting System by command line

Re: Too much of Forecasting Error due to input data

Re: SEASONAL ARIMA

Re: How to make forecast in report interactive?

Re: About Exponential smoothing time interval

Re: About Exponential smoothing time interval

Re: About Exponential smoothing time interval

Re: SAS VA and SAS Visual Statistics

Re: ADF test in Proc Arima

Re: Out-of-sample range vs holdout sample

Re: Too much of Forecasting Error due to input data

Re: Too much of Forecasting Error due to input data

Re: Time Series Forcasting System by command line

Re: Time Series Forcasting System by command line

Re: Unable to log on to SAS Forecast Environment

Re: Is there any way to move forecasting models produced by Forecast S...

Re: FORECAST Procedure, Winters method smoothing parameters

Follow Us

What is...