Programming the statistical procedures from SAS

Optimal way to use model previously fitted in HPLOGISTIC to score new dataset

Accepted Solution Solved
Reply
New Contributor
Posts: 4
Accepted Solution

Optimal way to use model previously fitted in HPLOGISTIC to score new dataset

[ Edited ]

I am using SAS 9.4. The documentation for HPLOGISTIC said that its focus were to fit and score large datasets. I am wondering what is the optimal to use it to

a) fit a model

b) then use the model to score some future datasets

 

I could think of at least two ways:

i) use OUTEST option then capture the parameter output with ODS, then use INEST with 0 iteration in a future LOGISTIC / HPLOGISTIC statement

ii) use the CODE option to generate the code for future scoring

 

Which one is more preferred? Or is there a better third way? Thank you!


Accepted Solutions
Solution
‎07-05-2017 01:02 PM
SAS Super FREQ
Posts: 3,753

Re: Optimal way to use model previously fitted in HPLOGISTIC to score new dataset

Posted in reply to clarkchong1

For an overview of the various methods, see "Techniques for scoring a regression model in SAS." Which you use is largely a matter of taste. The CODE method is slightly more complicated to deploy, but is often used on Big Data problems because you can the DATA step is so fast, or even run DS2 code in parallel on a grid.

View solution in original post


All Replies
Solution
‎07-05-2017 01:02 PM
SAS Super FREQ
Posts: 3,753

Re: Optimal way to use model previously fitted in HPLOGISTIC to score new dataset

Posted in reply to clarkchong1

For an overview of the various methods, see "Techniques for scoring a regression model in SAS." Which you use is largely a matter of taste. The CODE method is slightly more complicated to deploy, but is often used on Big Data problems because you can the DATA step is so fast, or even run DS2 code in parallel on a grid.

☑ This topic is solved.

Need further help from the community? Please ask a new question.

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