Graph with known intercept and slope

[Jie111]

I got some intercept and slopes from proc mixed effect model, like the following table show.

I want to make line graph using those intercept (25.00) and slope (0.040). How to do it in SAS?

[PaigeMiller]

PROC SGPLOT with the LINEPARM statement will do this. Maybe PROC PLM will also do this, I don't know, I never tried.

[Jie111]

Thank you for the reply. However, the LINEPARM must be used with another plot statement.

It is not possible to do like following:

proc sgplot;
lineparm x=0 y=26.8 slope=0.335;
run;

[PaigeMiller]

Then get creative — plot data that is transparent and can't be seen. Example:

proc sgplot data=sashelp.class;
scatter x=height y=weight/transparency=1;
lineparm x=0 y=26.8 slope=0.335;
xaxis min=0 max=200 label='Timepoint';
yaxis min=0 max=200 label='Y';
run;

Or look into using PROC PLM by saving the output of the PROC MIXED using the STORE command in PROC MIXED. https://documentation.sas.com/?cdcId=pgmmvacdc&cdcVersion=9.4&docsetId=statug&docsetTarget=statug_in...

[Jie111]

Interesting idea.

It works now.

Thanks for your help.

[Rick_SAS]

You can do this with PROC PLM without needing to hard-code the estimates. In your PROC MIXED call, add the STORE statement, such as

store out=MixStore;

Then call PROC PLM and use the EFFECTPLOT statement:

proc plm restore=MixStore;
   effectplot fit; /* or use the noclm option to suppress the CLM */;
run;

[Jie111]

Thanks for the reply. 

I did the graph using proc plm but it was not what I want. It only gave me one line representing the sex=1 but I want to show two lines for sex=0 and sex=1 (shown in the estimate table), respectively.

[PaigeMiller]

Show us the LOG for PROC PLM (not just any errors that may be in the log, but the entire log for the PROC PLM step).

Please click on the {i} icon and paste the log as text into that window.

[Jie111]

There is no error in the LOG. I attached with my code for mixed and plm.

                                %macro mix_group(dataset,group);
                             
                                   proc sort data=&dataset;
                                    by L_PNR;
                                   run;

                                   proc mixed data=&dataset;                                        
                                     class t &group;
                                     model col1=timepoint &group &group*timepoint /solution ddfm=kr;
                                     repeated t /subject=L_PNR type=un;
                                          estimate "intercept:&group=0" intercept 1 &group 0 1 /cl;
                                          estimate "intercept:&group=1" intercept 1 &group 1 0 /cl;
                                          estimate "intercept:1-0"   intercept 0 &group 1 -1 /cl;            
                                          estimate "slope for &group=0" timepoint 1  &group*timepoint 0 1  /cl;
                                          estimate "slope for &group=1" timepoint 1  &group*timepoint 1 0  /cl;
                                          estimate "slope for &group 1- &group 0" timepoint 0  &group*timepoint 1 -1  /cl; 

                                      store MixStore;
                                  run;
                                %mend;
                              
                                %mix_group(atl_two_bmi_3,sex);
                           


								 
proc plm restore=MixStore;
   effectplot fit;
run;

[Rick_SAS · Accepted solution]

OK. This is a different model that in your initial post, To display this model, use  the SLICEFIT(sliceby=sex) option. The doc for PROC PLM is here.

You might also want to read the article "Use the EFFECTPLOT statement to visualize regression models in SAS."

proc plm restore=MixStore;
   effectplot slicefit( sliceby=Sex );
run;

[Jie111]

Thanks. It is very helpful and powerful.