How to filter datasets in the enterprise miner

[badikidiki]

hello,

my enterprise miner project has the following nodes: sample, data partition and decision tree.

My sample consists of three groups A, B, C and I would like to run the analysis without group A.

In which node can I filter off group A or restrict the analysis to groups B and C?

Thanks for your help!

badikidiki

[M_Maldonado · Accepted solution]

Hi Badikidiki,

There are a few ways to do this. Try to reuse the SAS Code Node in this thread: https://communities.sas.com/thread/71517

I hope it helps!

Miguel

[badikidiki]

Hello Miguel,

this helps, I could filter my data and use it for further analysis.

Thanks a lot!

badikidiki

[BrettWujek]

Hi badikidiki - Glad you got your data filtered properly using the code Miguel pointed you to.  For what it's worth, it seems that you could have also just used the Filter node which is provided on the Sample tab in EM.  Simply hook up the Filter node to your input data node and in the properties for the Filter node click the "..." next to Class Variables.  In that dialog select the variable that you want to filter on and click on the appropriate value bars to filter out in the histogram (Ctrl-click to select multiple) and click Apply Filter.  When you run that node the output data set will exclude all observations with those values for that variable.

Give it a try if you get a chance.

Brett

[C_Ishihara]

You may also consider the Text Filter Node. You can filter on both fixed and unstructed variable data by using the Document Filter features.

Example: I want to find a word in a text field only relating to Product B. The snippet results are very helpful and can speed up review.