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

 

I am trying to build a model on an Interval target using SAS Enterprise Miner (v. 13.2) using Log transformation of the target and using the Linear Regression.

I have segmented the population based on different characteristics (like Low/High-income customers as model 1&2 and Male/Female customers as model 3&4) and ran a model for each segment.

Now trying to come up with the best segmentation based on the model outputs, I would like to use model comparison node to compare the fit statistics. the problem is, I cannot find a way to merge the segments together (1&2) vs. (3&4) so that I can compare the performance of the models together (Gender vs. Income level).

I used the "Append" node to merge two segments on each model but not successful.

 

I appreciate any help.

Sam

1 ACCEPTED SOLUTION

Accepted Solutions
JasonXin
SAS Employee
Hi, What you can do is to build a segment variable, say, segment_model by consolidating your segmentation logic. For example, if Gender = "F" and income="low", then segment_model=1; else if Gender = "F" and income="mid", then segment_model=2; else if Gender = "F" and income="high", then segment_model=3;... just build all the segmentation combinations into one variable. After reading the data set into EM, turn this new variable segment_model's role into Segment. Make sure there is only ONE variable acting as Segment Role. Then look into EM online example under START and END group. You should learn how to build models on this segment variable quickly. Then after END group you can connect Model Comparison Node. The Comparison node will report model statistics by the segments. You can always run the model based on the whole sample alongside this START-END group process. Hope this help? Jason Xin

View solution in original post

3 REPLIES 3
LinusH
Tourmaline | Level 20

Moved to the appropriate Data Mining forum.

Data never sleeps
JasonXin
SAS Employee
Hi, What you can do is to build a segment variable, say, segment_model by consolidating your segmentation logic. For example, if Gender = "F" and income="low", then segment_model=1; else if Gender = "F" and income="mid", then segment_model=2; else if Gender = "F" and income="high", then segment_model=3;... just build all the segmentation combinations into one variable. After reading the data set into EM, turn this new variable segment_model's role into Segment. Make sure there is only ONE variable acting as Segment Role. Then look into EM online example under START and END group. You should learn how to build models on this segment variable quickly. Then after END group you can connect Model Comparison Node. The Comparison node will report model statistics by the segments. You can always run the model based on the whole sample alongside this START-END group process. Hope this help? Jason Xin
Sam_zirak
Fluorite | Level 6

 

It works very well.

I cannot thank you enough for your guidance Jason.

Using start and eng groups is something I had never heard of before and it is incredibly powerful and time-saving. That is exactly what I was looking for and it will save me a huge amount of time on any project that I work on.

Thanks a million!

 

Sam

sas-innovate-2024.png

Join us for SAS Innovate April 16-19 at the Aria in Las Vegas. Bring the team and save big with our group pricing for a limited time only.

Pre-conference courses and tutorials are filling up fast and are always a sellout. Register today to reserve your seat.

 

Register now!

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
  • 3 replies
  • 3660 views
  • 1 like
  • 3 in conversation