Append Predicted Column in CART model

Hello, I am doing a Cart tree and I wonder if there is a way to append the predicted column that appears in the "Decision Tree Predictor" node to the whole base or data before doing the partition. 

I want to know because I think that I spend a lot of time applying and formulating the terminal nodes rules in Excel in order to find the predicted bin.  

Best Regards





I am not quite sure I understand the question.  Would it be possible for you to share an example workflow demonstrating what you are trying to accomplish?



Hi Aaron please find attached the example workflow




If I understand correctly, you can simply re-use the model output port with a new predictor node to evaluate all of your data with the model.  It is also possible to write the decision tree itself into a file and re-use it in another application, or even just a different workflow.  Attached you will find some examples of this. 

I would caution you though to be careful when using a model on the same data that you used to build it.  The fit may be better than you can expect in production as the model could be over-fit to the training data (if it is not a general solution). If evaulating the quality of your model is your concern at this point, I would look to cross validation.  This will let you re-run your model building process several times with the same settings to get a better idea for the stability of your model.  An example of this is also included. 



Thank you Aaron

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.