Community Spring Cleaning week is here! Join your fellow Maveryx in digging through your old posts and marking comments on them as solved. Learn more here!

Alteryx Designer Desktop Discussions

Find answers, ask questions, and share expertise about Alteryx Designer Desktop and Intelligence Suite.
SOLVED

Output the significant/used predictors from a tree model/random forest/boosted model/SVM

Koyel
6 - Meteoroid

Hi ,

 

Is there a way i can pull the used predictors in a tabular format from the predictive tools like decision tree / random forest / boosted model and SVM? I understand i need to use a R Script to get the information but cannot proceed further.

 

Any help will be highly appreciated.

 

Thanks,

Koyel

 

3 REPLIES 3
NickC
Alteryx Alumni (Retired)

Hello,

 

Sadly there is no easy way to extract these from the tools as they currently are. 

 

However if you right click on the tool and open the macro.  Search for the predictor variable interface tool and connect it up to a text input.  Add a Macro output and whatever gets selected will now come out of that tool as a separate anchor.  Save as a new macro and add to your existing workflow.

 

 

interface.png

 

Happy to discuss in more detail if needed.

 

Thanks,

Nick

 

 

yeongjeewei
5 - Atom

Hi Nick,

 

I try to follow your steps but there's an error, kindly advice.

 

Below are my steps:

1. right click on the predictive tool  

2. open macro

3. save as new macro 

4. add new macro to existing workflow (right click > insert > Macro)

5. new predictive tool will auto displayed (but when run got error as below)

 

 Linear Regression (39) Tool #107: Cannot find macro ".\Supporting_Macros\Linear_Model_summary.yxmc".
 Linear Regression (39) Tool #108: Cannot find macro ".\Supporting_Macros\Linear_Model_ANOVA.yxmc".

 

 

Thank you.

Jee

 

 

NickC
Alteryx Alumni (Retired)

Hello Jee,

 

If you right click on the linear regression and revert back to Linear Regression 1.0.

 

Then complete the same steps as you have done, you shouldn't have a problem. 

 

The linear regression tool has now got an HTML GUI, so the post I initially sent doesn't work exactly the same.

 

Nick

Labels