PROC HPSPLIT in SAS9.4

[tka726]

Hello!

I am trying to create a decision tree in SAS v9.4 (TS1M1) using PROC HPSPLIT. This is an entirely new procedure for me and it's a little daunting.  Any help is greatly appreciated!!

My outcome is a binary group, and I have a few binary predictors. The model will run, but the output is not what I expected. I get  line-by-line output, with nodes all out of order. I am struggling to make sense of it. I was expecting a nice decision tree. Is it because of the version of SAS I am using?

Also - is there a way to get sensitivity, specificity from this procedure?

Thanks so much!!!

PROC HPSPLIT DATA=ads plots=all maxdepth=7 maxbranch=2 ;
CLASS group race_white asa12 anes_type_gen surg_type male ;
MODEL group= race_white asa12 anes_type_gen surg_type male;
grow gini;
prune gini;
rules file='&path.\test.rtf';

run;

[Reeza]

Have you already tried the examples in the documentation?

I would suggest working through the examples in the documentation first and understand what's happening there and then working on your own data.

Most PROCs have a decent example section:

https://documentation.sas.com/?docsetId=statug&docsetVersion=15.1&docsetTarget=statug_hpsplit_exampl...

Code specifically:

https://documentation.sas.com/api/docsets/statug/15.1/content/statug_code_hpsplex1.htm?locale=en

---

@tka726 wrote:

Hello!

I am trying to create a decision tree in SAS v9.4 (TS1M1) using PROC HPSPLIT. This is an entirely new procedure for me and it's a little daunting.  Any help is greatly appreciated!!

My outcome is a binary group, and I have a few binary predictors. The model will run, but the output is not what I expected. I get  line-by-line output, with nodes all out of order. I am struggling to make sense of it. I was expecting a nice decision tree. Is it because of the version of SAS I am using?

Also - is there a way to get sensitivity, specificity from this procedure?

Thanks so much!!!

PROC HPSPLIT DATA=ads plots=all maxdepth=7 maxbranch=2 ;
CLASS group race_white asa12 anes_type_gen surg_type male ;
MODEL group= race_white asa12 anes_type_gen surg_type male;
grow gini;
prune gini;
rules file='&path.\test.rtf';

run;

---

[Reeza]

I've linked to the latest version of SAS/STAT but you should check the version you need. You can find it by running: 'proc product_status;run;' and it will be in the log - you're looking for SAS/STAT 14.3 or something like that.

[PGStats]

It is really worth upgrading to a more recent version of HPSPLIT. Especially for the graphs.If I recall correctly, the version you are using doesn't produce any graph. Check out the most recent doc to see what it does now.