Enterprise Miner - Additional variables

[elsolo21]

This may be a really basic question but I couldn't find the answer.  When Scoring a dataset, does it have to have ONLY the exact variables that the model uses?  With each model change, I keep changing the scoring dataset to match the best model.  If I leave all the original variables, will it just ignore the ones the model doesn't use?  At one point I saw that when scoring output from an HP Forest node it included the 'extra' variables in it's scoring 'scorecard' so I became a little confused.

Sorry if this is posted on the wrong board.  I didn't see a specific one for Enterprise Miner.

thanks!

[ballardw]

I think you answered your own question.

Since the purpose of scoring is basically to create a modeled value from the input variables a likely use would be compare an existing measurement or other modeled value with the current model result. Which would be difficult if the other measurement(s) were excluded.

[elsolo21]

Thanks for that response but I don't think I explained my question well enough.  This is a SAS EM 'logistics' question.  I already have a viable model and I'm scoring a separate dataset using that model.  However, If I need to tweak the model, the variables that are brought in could differ each time.  When that happens, I've been creating a new score dataset using only those new variables.  My question is can I have a scoring dataset will ALL the original variables?  Will the score node ignore the 'extra' variables not included in the final model node (in this case it's an HP Forest). This will speed up the process for evaluating the scoring node output.

thanks again.

[MikeStockstill · Accepted solution]

Hello Elsolo21-

The data set that you are going to use for scoring can contain variables that are not used by the score code.  You do not need to remove variables that are not used by the score code.

Have a great week.

[elsolo21]

Perfect! That's exactly what I was looking for. Thank You!