turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

Find a Community

- Home
- /
- SAS Programming
- /
- SAS Procedures
- /
- Logit model

Topic Options

- RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

03-18-2009 11:08 AM

I am having trouble running an appropriate logit model. The final goal is to estimate the probabilities of each horse to win the race. I can provide sample data trough email but it is somewhat large to put on here, but I may be able to do so if needed.

The data is set up so that each horse and its characteristics are contained in one row (one observation). The identifying variables in each row are track, date and race number. I would like to run a model that will estimate probabilities by race number and these probabilities will sum to unity. The probability of each horse winning also depends on the other horses in the field.

I can run a simple proc logistic but the probability estimates from this model will not sum to one. There may be other ways to do this, proc catmod, or proc MDC.

Finally a simpler question: I would like to input data into the model after I make it to produce probability estimates for test data that was not used in the model creation. How would I do this. Even something simple like a linear regression with 10 covariates. Just something to run data trough the estimated formula.

Please let me know if you can help. I can provide more details and data if required. Thank you.

The data is set up so that each horse and its characteristics are contained in one row (one observation). The identifying variables in each row are track, date and race number. I would like to run a model that will estimate probabilities by race number and these probabilities will sum to unity. The probability of each horse winning also depends on the other horses in the field.

I can run a simple proc logistic but the probability estimates from this model will not sum to one. There may be other ways to do this, proc catmod, or proc MDC.

Finally a simpler question: I would like to input data into the model after I make it to produce probability estimates for test data that was not used in the model creation. How would I do this. Even something simple like a linear regression with 10 covariates. Just something to run data trough the estimated formula.

Please let me know if you can help. I can provide more details and data if required. Thank you.

- Mark as New
- Bookmark
- Subscribe
- RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to deleted_user

03-18-2009 11:48 AM

Regarding your final question, run the following to get predicted test result.

proc logistic data = model_data outmodel = logit_model;

model event(event = '1') = x;

run;

proc logistic inmodel= logit_model;

score data = test_data out= scored_test_data;

run;

proc logistic data = model_data outmodel = logit_model;

model event(event = '1') = x;

run;

proc logistic inmodel= logit_model;

score data = test_data out= scored_test_data;

run;

- Mark as New
- Bookmark
- Subscribe
- RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

03-18-2009 03:36 PM

awesome, thank you.

I still can't figure out how to specify the model correctly though. If anyone has any thoughts please let me know.

I still can't figure out how to specify the model correctly though. If anyone has any thoughts please let me know.