topic Re: How to Export Logits in Enterprise Miner in SAS Data Science

How to Export Logits in Enterprise Miner

andreas_zaras — Thu, 21 Mar 2013 20:05:41 GMT

Hello,

I am using the utility bode for custom code in EM to export a scored data set form a process flow. The variables that i get from the best model (logist regression) are ok (e.g. probabilities of primary and secondary event, predicted target etc) but i don;t get the logit (log odds) from the regression model. How can i get this variable also?

Thnaks in adavnce,

Andreas

Re: How to Export Logits in Enterprise Miner

adjgiulio — Thu, 21 Mar 2013 22:33:42 GMT

I'm interested in seeing if someone can answer your questions. I've looked into this same question myself not long ago. A logistic regression in EM is run through PROC DMREG. When you look at DMREG's documentation it doesn't say anything about an option to export coeffiecients. The workaround I found, which is no good if you need an automated solution but worked for what I had to do that one time, is the following:

-go to the results of your regression.

-by default EM shows an Effect Plots chart. Highlight that chart.

-now go to View->Table. This will open up a table that has a few pieces of information on your inputs, including their coefficients which you can use to calculate odds.

-you can now select all the data in the table and paste it in Excel, or you can save the table as a SAS table (File->Save as, after having highlighted the table)

Re: How to Export Logits in Enterprise Miner

andreas_zaras — Thu, 21 Mar 2013 23:14:33 GMT

Thanks for your answer. This can be a solution but there must be an option to produce automatically the logits. EM produces so much output that i cannot imagine that it can;t produce the logits. I hope someone can answer. A turnaround that i just thought is to produce the log odds (logits) using the sas code node in the utility tab. Since the probability of the primary event is reported i guess we can produce it by a data step using an assignment statement to produce log(prob_b1/(1-prob_b1)). The expression can be assigned to a variable called logit.

Let us see if the automatic way is done!

Thnaks again