BookmarkSubscribeRSS Feed
znhnm
Quartz | Level 8

Hello,

 

I have a logistic regression model where I do the scoring using the Export model... option, getting the code to build the model on VA and then putting it on SAS Studio making some minor changes to add the scoring data and I finally complete the scoring. 

znhnm_0-1684852647759.png


Also, the default prediction cutoff is 0.5 in VA and it is indicated in the code with this line: 

if P_StatusDead >= 0.5

znhnm_1-1684853224175.png

 



However, I realize that when I change the Prediction cutoff on VA, say, make it 0.3, the code doesn't get updated. Why this is happenning and is there a way to prevent from this to happen? 

Thank you !

1 REPLY 1
acordes
Rhodochrosite | Level 12
The threshold only affects the label on which the probability is categorized like event yes or no.

And therefore it affects accuracy metrics confusion matrix and false positives.

The formula for the logistic regression is independent from the threshold value.

As you treat the scoring code in SAS Studio you can easily change the value in the code.

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!

Tips for filtering data sources in SAS Visual Analytics

See how to use one filter for multiple data sources by mapping your data from SAS’ Alexandria McCall.

Find more tutorials on the SAS Users YouTube channel.

Discussion stats
  • 1 reply
  • 374 views
  • 0 likes
  • 2 in conversation