Note, you do need SAS 9.2 or later version to use SG procedures. ... View more

While the above graph looked reasonable right after I made it, when I came back to it, I could not understand why it appeared as if Hybrids had an average weight of over 4000 lbs.  The association between the left axis and the bar chart is very strong, that any consumer of this graph may be mislead to think the bar represents weight, even when it is labeled with the Mileage value.  If we want only one axis, the following graph with both plots showing data labels may be better: proc means data=sashelp.cars;   class type;   var mpg_city weight;   output out=carmeans mean=mpg_mean wt_mean;   run; data cars;   set carmeans;   length wt_label $16;   wt_label=put(wt_mean, 5.0) || ' lbs';   run; proc print;run; proc template;   define statgraph cars;     begingraph;       entrytitle 'Mean Mileage and Weight by Type';       layout overlay / y2axisopts=(display=none offsetmin=0 linearopts=(viewmin=0))                        yaxisopts=(offsetmin=0 linearopts=(viewmin=0))                        xaxisopts=(display=(ticks tickvalues line));         barchart x=type y=mpg_mean /  barlabel=true name='m' barlabelattrs=(size=9);         seriesplot x=type y=wt_mean / yaxis=y2 name='w' datalabel=wt_label                    lineattrs=(thickness=3) display=(markers) markerattrs=(symbol=circlefilled) ;       discretelegend 'm' 'w';       endlayout;     endgraph;   end; run; ods listing gpath='C:\Work\Blogs\Communities\images'; ods graphics / reset width=5in height=3in imagename='CarMpgWeight2'; proc sgrender data=cars template=cars;   format mpg_mean 4.1;   format wt_mean 5.0; run; ... View more

The simplest solution is: proc sgplot data=sashelp.cars;   vbar type / response=mpg_city stat=mean;   vline type / response=mpg_highway stat=mean;   run; The data range for the Y axis will be a union of the two. SGPLOT VBAR statement does not support Y2AXIS.  You can use GTL to do everything you want provided you summarize the data for the SERIESPLOT.  I will add an example shortly. ... View more

You can do this using SGPLOT procedure.  If this is what you need, I can attach the code. ... View more

Yes, with SAS 9.3 (see options mentioned).  Not with SAS 9.2  SG procs or GTL.  You could use proc gchart and annotate. ... View more

Really easy with SAS 9.3 where you can just use X=Month, GROUP=Product and GROUPDISPLAY=Cluster. But I assume you are using SAS 9.2.  If you have the M3 maintenence release, you can structure you data as multi-variable, and then overlay the bars as needed.  You will need to customize your program and data based on how many groups you have.  See example below with three products. data MultiErrorMonth;   input Month $  Revenue1 Upper1 Lower1 Revenue2 Upper2 Lower2 Revenue3 Upper3 Lower3;   datalines; Jan  2000  2200  1800 1500 1700  1300  1200 1300 1100  Feb  2200  2400  1900 1200 1400  1300  1300 1400 1100  ; run; ods listing; proc print;run; proc template;   define statgraph MultiError;     begingraph;       entrytitle 'Clustered Bars with Confidence Intervals';       layout overlay /  xaxisopts=(offsetmin=0.25 offsetmax=0.25 display=(ticks tickvalues))                         yaxisopts=(label='Revenue' griddisplay=on) cycleattrs=true;          barchartparm x=month y=revenue1 / errorupper=upper1 errorlower=lower1 skin=modern                      discreteoffset=-0.3 barwidth=0.3 name='Desks' legendlabel='Desks';          barchartparm x=month y=revenue2 / errorupper=upper2 errorlower=lower2 skin=modern                      discreteoffset=-0.0 barwidth=0.3 name='Chairs' legendlabel='Chairs';          barchartparm x=month y=revenue3 / errorupper=upper3 errorlower=lower3 skin=modern                      discreteoffset=0.3 barwidth=0.3 name='Lamps' legendlabel='Lamps';           discretelegend 'Desks' 'Chairs' 'Lamps';        endlayout;      endgraph;   end; run; ods listing; ods graphics / reset width=500px height=300px imagename='MultiError'; proc sgrender data=MultiErrorMonth template=MultiError; run; ... View more