[Ankit Bhardwaj]

I have run following commands:

* missing values- mdesc

* working with a subset that is empty in your data- mdesc

* supplying a string variable where a numeric variable is needed- varmanage

[Nick Cox]

I have never used this program, but your call restricts observations to those where the second variable is a constant. Perhaps that is an infeasible request. I am still speculating.

Without more information, I won't be able to say more. Perhaps you should write your own program to do what you want.

[Joie Ensor]

Dear Statisticians,
Kindly help me with the calibration plot. I am not able to get it after running the pmcalplot command; instead, I am getting the below error message.

pmcalplot p_dev VNTSept if VNTSept ==1, ci
Binary option selected: Calibration plot for logistic prediction model displaying...
r(2000);

Hi Ankit Bhardwaj,

Sorry to hear you're having trouble with pmcalplot. Nick contacted me via email to see if I can help. Hopefully we can resolve your issue.

If we look back at your command line in post #12 above, while the format looks correct, the logic causes an issue. p_dev is a variable containing the predicted probabilties from your model. VNTSept should be your outcome variable, coded [0,1] with non-events==0 and events==1.

I suspect if you remove the if statement, the command will execute without error and produce your plot. The if statement here limits the calculations to the subgroup of events only. This would make little sense for creating a calibration plot, where we are looking to assess how well a models predictions match observed outcomes.

The 'if' clause can be used with pmcalplot, but it would more commonly be used to plot for say only the training set, or only the test set for example. There is an example like this in the help file. Perhaps that is where you have derived your code including the 'if' from?

Please let me know if this helps.

And thank you to all those who have tried to help earlier in the thread.

Best wishes,
Joie