PROC UCM low bias back transformation

[JustABitOutside]

I am running PROC UCM to forecast ln(sales). The log() function is being used in a preceding DATA step to transform the series. In order to transform back to the original scale and get forecasts of the mean (and not the median), I need the forecasts on the log scale and the estimated residual variance. What's the easiest way (or any way) in PROC UCM to output the residual variance from the model? Thank you in advance for any input you are able to provide. 

[rselukar · Accepted solution]

You can specify an OUTFOR= data set in the FORECAST statement.  The STD column in this data set is what you need.

[JustABitOutside]

Thank you! This is exactly what I needed.