This is a cumulative logit model. When combined with the spline effects, I think it will be hard to obtain an analytical expression for the derivative of the predicted responses. It's much easier to use a numerical approximation. I propose:   Create scoring data that uses small steps between points in the independent variable (HEIGHT). Use the DIF function to form the finite-difference derivatives.  See also the article, "Compute derivatives for nonparametric regression models." Here is the code that creates the scoring data and that visualizes the predicted responses. You can use the links I've provided to obtain the finite-difference derivatives at any (sex, height) combination.   proc means data=sashelp.class; class sex; var height; run; data scoreData; sex='F'; do height = 51.4 to 66.5 by 0.1; output; end; sex = 'M'; do height = 57.5 to 72 by 0.1; output; end; run; proc logistic data=sashelp.class; class sex(ref="F")/param=ref; effect htS=spline(height/ degree=1 knotmethod=percentilelist(25,75)); model age=htS sex htS*sex; score data=scoreData out=ScoreWide; run; title "Visualize the Cumulative Logit Model"; proc sgpanel data=ScoreWide; panelby sex; series x=height y=P_11; series x=height y=P_12; series x=height y=P_13; series x=height y=P_14; series x=height y=P_15; series x=height y=P_16; run;   ... View more

Is there anything else we can help you with? If not, please accept a solution and close this thread. ... View more

No. PROC REG assumes that the dependent variable (the response) is continuous. If your dependent variable is binary, you should probably switch to using PROC LOGISTIC, which (by default) models the probability that a response is 0. ... View more

I read the documentation. I think it says the following: The procedure fits a cubic spline to the data and evaluates the spline for every day. If you use the option observed=(beginning), that is equivalent to using observed=(beginning,beginning).  That means that the output value is the value of the cubic spline on the first day of each month: 01JAN, 01FEB, 01MAR, etc. If you use the option  observed=(beginning,average), the output value is the AVERAGE of the values in the month. Thus the output for January is the average of the cubic spline in January, the output for February is the average of the cubic spline in February, etc. Here is the graph of the cubic interpolation function (which I got by setting TO=DAY and PLOTS=(ALL)). See if this helps you make sense of the outputs in both cases:   ... View more

PROC REG does not support a CLASS statement, so there is no default reference level.  When using PROC REG, you have to create the dummy variables yourself.   Let's use the example of creating a dummy variable for a two-level variable such as GENDER. Your reference level is always the lowest level, which is 0 if you construct a 0/1 binary variable.  For example, if you define sex = (Gender='M'); then the '0' level corresponds to females.  However, if you define  sex = (Gender='F'); then the '0' level corresponds to males.    That is why I recommend including the reference level in the variable name, such as  isMale = (Gender='M');       ... View more

If your question is answered, please close this thread. If you have additional questions, please let us know how we can help. ... View more

I feel confident that you can use the best practices in the links to improve your program. Since I don't have your data, it is difficult for me (or anyone) to know whether our suggestions fix all of the problems you are experiencing. Give it a try and then report back if you have additional questions.  ... View more

I don't want to hijack this thread, but I'd like to encourage SAS Experts like @Kathryn_SAS and @Quentin to stop proposing  ODS _ALL_ CLOSE; or  ODS LISTING CLOSE; in their solutions.  The modern syntax for suppressing ODS results is ODS EXCLUDE ALL; followed by ODS EXCLUDE NONE or ODS SELECT <object> when you are ready to resume output. For more details, see  What is the best way to suppress ODS output in SAS? Five reasons to use ODS EXCLUDE to suppress SAS output ... View more

The histogram bin widths (and therefore the number of bins) are not only determined by n, the number of nonmissing values, but also by choosing bin widths that are "convenient", as described in Lewart (Algorithm 463 of the Collected Algorithms of the ACM, 1973). You can get the bin locations from Lewart's algorithm by using the GSCALE subroutine in SAS IML.  For detail, examples, and a discussion, see The location of ticks in statistical graphics - The DO Loop ... View more

This is all conjecture, but I think you'll find that the predictions from both models are essentially equivalent. It is also likely the differences between the predicted values from these models are smaller than the errors in the data.   Recall that the parameter estimates are the solution of a maximum likelihood optimization. When we optimize a function, we iterate until some criteria are met. If you add the ITPRINT option to the MODEL statement, you can compare the optimization path used by the old and new version of SAS. My guess is that you will see a very similar path, with minor differences explained by the newer chips on the new server.  ... View more

OK. Can you post the before/after estimates? I'd like to know if they are close (for example, differ in the 4th decimal place) or are vastly different.   I assume the server that SAS runs on has also changed? For example, you used to run on a Windows Workstation but now you are running on a Linux server on AWS in the cloud? ... View more

Warren gave you part of the solution: it is best practice to put ODS statements inside the procedure.   In addition, you should remove your ODS LISTING CLOSE statements and replace them with ODS EXCLUDE ALL statements. Then insert ODS EXCLUDE NONE statements after you are done suppressing the SAS output.     ... View more

>  we get different output in Viya than we did in Enterprise Guide  It sounds like you moved from SAS 9.4 to SAS Viya, is that correct? Can you describe the differences (or post images of them)? Are the differences in tables? Graphs? The OUTEST= data set? If the difference appears in tables, use ODS TRACE ON to discover the names of the tables. Do you see the differences for a large variety of data sets or only for one data set?     ... View more