Proc sgpanel: Display R-square in each panels w/layout=lattice option

[Cruise]

Hi folks,

I have couple questions on the same plotting problem.

1. How to display r-square value in each panel? Insert, commented out now, works without lattice layout. But gives me an error "there is no variable R_square" after I add layout=lattice option. 

2. The bottom column's intention is to show relationship of "N of cases vs N of exposure sites". Because both cases and the N of exposure sites are sparse, the plot is not being very infortive. Any alternatives to recommend to visualize this context better?

Please note that this data is only 1% random sample of my actual data.

Thanks in advance. Let me know if you suggest me to seperate this post to two for each questions asked here.

I greatly appreciate any suggestions and hints or direct help.

DATA HAVE;
INPUT GEO pover $ N EXPOSURES $ EJ_VALUES;
CARDS;
36005039 4 2 CANCER_RISK 63.2836 
36005039 4 2 COUNTS_TSDF 0.0000 
36007011 3 0 CANCER_RISK 22.3069 
36007011 3 0 COUNTS_TSDF 0.0000 
36009960 2 0 CANCER_RISK 18.1951 
36009960 2 0 COUNTS_TSDF 1.0000 
36017970 4 0 CANCER_RISK 38.6816 
36017970 4 0 COUNTS_TSDF 1.0000 
36027190 2 0 CANCER_RISK 32.9410 
36027190 2 0 COUNTS_TSDF 0.0000 
36029006 3 0 CANCER_RISK 35.1154 
36029006 3 0 COUNTS_TSDF 0.0000 
36029010 3 0 CANCER_RISK 31.6126 
36029010 3 0 COUNTS_TSDF 0.0000 
36039080 3 0 CANCER_RISK 20.8220 
36039080 3 0 COUNTS_TSDF 0.0000 
36047000 3 0 CANCER_RISK 70.9754 
36047000 3 0 COUNTS_TSDF 2.0000 
36047004 2 0 CANCER_RISK 63.1364 
36047004 2 0 COUNTS_TSDF 4.0000 
36047008 4 0 CANCER_RISK 56.5344 
36047008 4 0 COUNTS_TSDF 0.0000 
36047011 3 0 CANCER_RISK 62.7163 
36047011 3 0 COUNTS_TSDF 0.0000 
36047044 4 1 CANCER_RISK 57.5470 
36047044 4 1 COUNTS_TSDF 3.0000 
36047048 4 1 CANCER_RISK 52.8631 
36047048 4 1 COUNTS_TSDF 0.0000 
36051030 3 0 CANCER_RISK 20.6163 
36051030 3 0 COUNTS_TSDF 0.0000 
36059404 1 1 CANCER_RISK 45.0740 
36059404 1 1 COUNTS_TSDF 0.0000 
36059518 1 0 CANCER_RISK 46.2042 
36059518 1 0 COUNTS_TSDF 1.0000 
36061002 3 1 CANCER_RISK 75.1446 
36061002 3 1 COUNTS_TSDF 0.0000 
36061011 2 0 CANCER_RISK 87.4068 
36061011 2 0 COUNTS_TSDF 2.0000 
36061018 2 1 CANCER_RISK 82.4592 
36061018 2 1 COUNTS_TSDF 1.0000 
36061023 3 1 CANCER_RISK 65.9496 
36061023 3 1 COUNTS_TSDF 0.0000 
36063024 3 0 CANCER_RISK 22.5988 
36063024 3 0 COUNTS_TSDF 2.0000 
36063024 3 0 CANCER_RISK 22.7899 
36063024 3 0 COUNTS_TSDF 0.0000 
36067011 2 0 CANCER_RISK 33.3920 
36067011 2 0 COUNTS_TSDF 0.0000 
36067015 2 0 CANCER_RISK 28.8998 
36067015 2 0 COUNTS_TSDF 0.0000 
36069051 2 1 CANCER_RISK 20.2576 
36069051 2 1 COUNTS_TSDF 0.0000 
36075021 3 0 CANCER_RISK 21.6302 
36075021 3 0 COUNTS_TSDF 0.0000 
36077591 2 0 CANCER_RISK 19.2683 
36077591 2 0 COUNTS_TSDF 0.0000 
36079011 2 0 CANCER_RISK 30.6732 
36079011 2 0 COUNTS_TSDF 1.0000 
36081040 4 2 CANCER_RISK 63.5312 
36081040 4 2 COUNTS_TSDF 0.0000 
36081086 4 1 CANCER_RISK 62.9083 
36081086 4 1 COUNTS_TSDF 0.0000 
36081098 1 0 CANCER_RISK 46.6523 
36081098 1 0 COUNTS_TSDF 1.0000 
36081117 4 0 CANCER_RISK 60.2166 
36081117 4 0 COUNTS_TSDF 1.0000 
36081162 3 0 CANCER_RISK 55.5270 
36081162 3 0 COUNTS_TSDF 1.0000 
36083040 4 0 CANCER_RISK 47.1551 
36083040 4 0 COUNTS_TSDF 1.0000 
36085022 1 0 CANCER_RISK 45.1212 
36085022 1 0 COUNTS_TSDF 1.0000 
36087010 2 0 CANCER_RISK 33.8989 
36087010 2 0 COUNTS_TSDF 0.0000 
36089491 3 0 CANCER_RISK 24.4336 
36089491 3 0 COUNTS_TSDF 1.0000 
36089492 3 1 CANCER_RISK 16.8045 
36089492 3 1 COUNTS_TSDF 0.0000 
36093020 1 0 CANCER_RISK 34.4799 
36093020 1 0 COUNTS_TSDF 0.0000 
36103111 2 1 CANCER_RISK 34.6693 
36103111 2 1 COUNTS_TSDF 0.0000 
36103159 3 0 CANCER_RISK 25.0552 
36103159 3 0 COUNTS_TSDF 0.0000 
36117020 1 0 CANCER_RISK 21.9895 
36117020 1 0 COUNTS_TSDF 0.0000 
36119000 4 0 CANCER_RISK 60.1986 
36119000 4 0 COUNTS_TSDF 0.0000 
36119004 2 0 CANCER_RISK 59.4253 
36119004 2 0 COUNTS_TSDF 0.0000 
36119013 1 0 CANCER_RISK 34.7585 
36119013 1 0 COUNTS_TSDF 1.0000 
;

PROC FORMAT;
value $POVER
'1'="POVERTY <5%"
'2'="POVERTY 5% to <10%"
'3'="POVERTY 10% to < 20%" 
'4'="POVERTY >20%"
'99'="Unknown"
;
run;

ods graphics / height=400px width=800px;
TITLE "AAR VS ESJCREEN";
proc sgpanel data=HAVE;
panelby POVER EXPOSURES /onepanel LAYOUT=LATTICE novarname ROWS=2;
REG x=EJ_VALUES y=N/ CLI CLM markerattrs=(size=3px);
FORMAT POVER $POVER.;
/*INSET R_Square;*/
keylegend/  title="MYTITLE" position=bottom;
colaxis label='NAME' fitpolicy=thin valuesformat=best4.0;
rowaxis label='LABEL' grid;
title "NIKE"; 
WHERE POVER NOT IN ("99") OR MISSING(POVER);
run;

[Cruise · Accepted solution]

I solved the problem, my own lengthy and time consuming way.If nobody posts easier solution then below might give an idea that I calculated the R-square separatelt and merged to my data for plotting. The code is not intended to be exact but just for an idea that INSET recalls variables from your plot data at least when it comes to R-square for your regressions. 

PROC SORT DATA=E.LONG_LUNG;
BY POVER EXPOSURES;
RUN;
ods graphics off; 
ODS OUTPUT FitStatistics=FIT;
PROC REG DATA=E.LONG_LUNG;
BY POVER EXPOSURES NOTSORTED;
MODEL AAR=EJ_VALUES;
ODS TRACE ON;
WHERE POVER NE '99';
RUN;

DATA RSQUARE_LUNG(RENAME=(cValue2=R_Square));
SET FIT(KEEP=EXPOSURES DEPENDENT LABEL2 cValue2 POVER);
WHERE LABEL2 IN ("R-Square");
RUN;

DATA E.LONG_LUNG1;
MERGE E.LONG_LUNG
      RSQUARE_LUNG;
BY POVER EXPOSURES ;
RUN;
      
ods graphics / height=500px width=1000px;
proc sgpanel data=E.Long_LUNG1;
panelby POVER EXPOSURES /onepanel LAYOUT=LATTICE novarname ROWS=2;
REG x=AAR y=EJ_VALUES/ CLI CLM markerattrs=(size=3px);
FORMAT POVER $POVER.;
INSET R_Square;
keylegend/  title="Age Adjusted Rate, per 100K" position=bottom;
colaxis label='NAME' fitpolicy=thin valuesformat=best4.0;
rowaxis label='EJSCREEN VARIABLES' grid;
TITLE "Lung and Bronchus, Age Adjusted Rate vs Select EJSCREEN Variables";
WHERE POVER NOT IN ("99") OR MISSING(POVER);
run;