Forecasting using SAS Forecast Server, SAS/ETS, and more

Forecast Studio - applying custom transformation of dependent variable?

Accepted Solution Solved
Reply
Occasional Contributor
Posts: 9
Accepted Solution

Forecast Studio - applying custom transformation of dependent variable?

I'd like to apply a custom transformation (for example, the inverse hyperbolic sine function) to the dependent variable before the models are run to generate forecasts in Forecast Studio. 

 

However, I'd also like to be able to see the usual graphs / forecasts etc in Studio showing the untransformed variable (or re-transformed back to the original scale) variable.

 

Is there a way to do this?

 

Right now, it looks like the transformations available in Studio don't allow for any customization (you either take whatever's provided by Studio -- E.g. Box-Cox, log etc -- or nothing at all)?

 

I can of course go ahead and load a dataset with the transformed dependent variable into Studio but then all the graphs etc generated will just be based on that and there's no way to see them on the original scale.

 

I am somewhat familiar with Studio's batch code as well -- so, if there's any way to do it through that, I'd greatly appreciate any pointers. 

 

Thanks!


Accepted Solutions
Solution
‎06-28-2017 03:27 PM
SAS Employee
Posts: 38

Re: Forecast Studio - applying custom transformation of dependent variable?

I can't think of a way to do that in the UI. If you provide a data set with transformed variables you can use a report to display the untransformed data. However you have to write the code for that. 

 

View solution in original post


All Replies
Solution
‎06-28-2017 03:27 PM
SAS Employee
Posts: 38

Re: Forecast Studio - applying custom transformation of dependent variable?

I can't think of a way to do that in the UI. If you provide a data set with transformed variables you can use a report to display the untransformed data. However you have to write the code for that. 

 

☑ This topic is solved.

Need further help from the community? Please ask a new question.

Discussion stats
  • 1 reply
  • 130 views
  • 1 like
  • 2 in conversation