BookmarkSubscribeRSS Feed
🔒 This topic is solved and locked. Need further help from the community? Please sign in and ask a new question.
ashtad63
Fluorite | Level 6

I am using HP Forst node in SAS Enterprise Miner 14.2 to select the top X (i.e. the top 10) important input variables among hundreds of input variables to predict a binary target variable. To do so, in the property window of HP Forest node, I selected "Yes" for Variable Selection option. Then, to specify how many variables to select, you should select "Random Branch Assignments (RBA)" as the Variable Importance Method. Once you do this selection, SAS EM gives you the possibility to manually enter the Number of Variables to Consider (i.e. pick the top 10 important variables).

 

SAS EM Help says the "RBA Margin Reduction" should be considered as a measure of variable importance when you have a class target variable (i.e. binary target variable). As expected, in the HP Forest node output window and in the Variable Importance table, you will see the RBA Margin Reduction column. I have attached an example Variable Importance table in which I have asked SAS EM to select the top 22 important variables in my data. My problem is that I don't know how the numbers in the RBA Margin Reduction column are calculated. What is the equation (or procedure) for calculating RBA Margin Reduction?

 

The answer may be found in the article below, but unfortunately, I couldn't get access to the article.

 

Neville, P. G., and Tan, P.-Y. (2014). “A Forest Measure of Variable Importance Resistant to Correlations.” In Proceedings of the 2014 Joint Statistical Meetings. Alexandria, VA: American Statistical Association.

 

I appreciate if you help me with my question.  

 

A. J. 

1 ACCEPTED SOLUTION

Accepted Solutions
PadraicGNeville
SAS Employee

Yes Ashtad63, you are correct:

"calculate Margin Reduction once without RBA and another time with RBA and the difference are the RBA Margin Reduction numbers appear in Variable Importance table output"

-Padraic

 

View solution in original post

4 REPLIES 4
ashtad63
Fluorite | Level 6

Thanks, Wendy for the link. I have seen this information before in SAS Enterprise Miner 14.3 High-Performance Procedures manual (there is more information in that manual as well). But my problem is that I don't know how RBA Margin Reduction in SAS variable Importance table is calculated.

 

I guess for calculating RBA Margin Reduction we should calculate Margin Reduction once without RBA and another time with RBA and the difference are the RBA Margin Reduction numbers appear in Variable Importance table output which I attached its screenshot in my last post. However, I am not sure about this. I may be wrong.

 

 

 

 

PadraicGNeville
SAS Employee

Yes Ashtad63, you are correct:

"calculate Margin Reduction once without RBA and another time with RBA and the difference are the RBA Margin Reduction numbers appear in Variable Importance table output"

-Padraic

 

ashtad63
Fluorite | Level 6

Thanks, Mr. Neville for confirmation. 

SAS Innovate 2025: Save the Date

 SAS Innovate 2025 is scheduled for May 6-9 in Orlando, FL. Sign up to be first to learn about the agenda and registration!

Save the date!

How to choose a machine learning algorithm

Use this tutorial as a handy guide to weigh the pros and cons of these commonly used machine learning algorithms.

Find more tutorials on the SAS Users YouTube channel.

Discussion stats
  • 4 replies
  • 1452 views
  • 0 likes
  • 3 in conversation