SAS Proc GCHART equivalent of Proc GPLOT UNIFORM

Question

Is there a way to make multiple bar charts with uniform axis with proc gchart?

In proc gplot, I can use the uniform option like this:

proc gplot data=test uniform;
  by state;
  plot var*date;
run;

This will give me a set of plots for the by variable that all use the same axis range.

This option doesn't exist for proc gchart--is there any other way to do this? I can't just define a fixed range since my data will vary.

What version of SAS do you have? If 9.2+, are you averse to SGPLOT/SGPANEL ? — Joe
I am moderately adverse to SGPLOT--this is using 9.2 which doesn't get nice vector output to PDF from SGPLOT. I can access 9.3 to run the program (and get clean SGPLOT output), but I'd prefer to make it work in 9.2 — orh
And this is going into an ODS Layout with some proc GPLOTS so I was trying to keep it all in the same family (never used sgplot with annotation and ODS layout). — orh
Tried an SGPLOT, and I could get uniform axis, but I can't make it work with GREPLAY (which I am using to set up my layouts), so I suspect SG* functions are out. — orh
Hmm, GREPLAY wouldn't work with SGPLOT. If you have 9.3, I suggest in the long run learning GTL, which will let you do a lot of this stuff more easily (including taking over the GREPLAY functionality to define layouts, and taking over uniformity of axis and such). — Joe

orh orh · Accepted Answer · 2013-12-06T16:32:35

Thanks for the input everyone. Since it looks like there isn't a good solution within the proc itself, I went with a macro approach to manually setting the axis.

This paper provided the foundation for what I did: http://analytics.ncsu.edu/sesug/2012/BB-09.pdf

Since I couldn't find the text of the program anywhere except for in that non-searchable PDF, I've typed it in here. My version adds one additional parameter that optionally pads the low value in order to leave space for data labels below the low point (useful if you are making a column chart with labels above the positive values and below the negative values)

%macro set_axis_minmaxincrement(ds=,
                                axisvar=,
                                axis_length = 51,
                                sa_min = 999999,
                                sa_max = -999999,
                                returned_min = axis_min,
                                returned_max = axis_max,
                                returned_increment = axis_increment,
                                force_zero = 0,
                                pad_bottom = 0
                                ) ;

%global &returned_min &returned_max &returned_increment;

/* Find the high and low values. Note: a data step was used versus a proc */
/* to allow the application of the option parameters, if specified. */
proc sort data=&ds out=sortlb(keep=&axisvar);
  by &axisvar;
  where &axisvar ne .;
run;
data axisdata(keep=low high);
  retain low 0;
  set sortlb end=eof;
  by &axisvar;
  if _n_=1 then low = &axisvar;
  if eof then do;
    high = &axisvar;
    if &sa_min ^= 999999 and &sa_min < low then low = &sa_min;
    if &sa_max ^= -999999 and &sa_max > high then high = &sa_max;
    %if &force_zero = 1 %then %do;
      if low > 0 then low = 0;
      else if high < 0 then high = 0;
    %end;
    %if &pad_bottom = 1 %then %do;
      if low < 0 then low = low-((high-low)*.06);
    %end;
    output;
  end;
run;

data axisdata;
  set axisdata;
  /* insure that high is greater than low */
  if high <= low then do;
    if abs(low) <= 1 then high = low + 1;
    else high = low+10;
  end;

/* Calculate the conversion unit to transform the standard range to */
/* include the actual range. This value is used to convert the standard */
/* to the actual increment for the actual range. */

  axisrange = high - low;
/* ranges of less than 1 */
  if axisrange <= 6 then do;
    check = 6;
    conversion_unit = .01;
    do until (axisrange > check);
      check = check/10;
      if axisrange <= check then conversion_unit = conversion_unit / 10;
    end;
  end;

/* Ranges of 1 or greater */
  else do;
    check = 60;
    conversion_unit = 1.0;
    do while (axisrange > check);
      check = check*10;
      conversion_unit = conversion_unit * 10;
    end;
  end;
/* standardize the range to lie between 6 to 60 */
    unit_range = axisrange/conversion_unit;
/* Set the increment based on the unitized range */
/* 'Long' axis, 8 - 12 increments */
    %if &axis_length >50 %then %do;
      if      unit_range < 12 then axisinc = 1   * conversion_unit;
      else if unit_range < 24 then axisinc = 2   * conversion_unit;
      else if unit_range < 30 then axisinc = 2.5 * conversion_unit;
      else                         axisinc = 5   * conversion_unit;
    %end;
/* Otherwise, 'short' axis, 4-6 increments */
    %else %do;
      if      unit_range < 12 then axisinc = 2   * conversion_unit;
      else if unit_range < 18 then axisinc = 3   * conversion_unit;
      else if unit_range < 24 then axisinc = 4   * conversion_unit;
      else if unit_range < 30 then axisinc = 5   * conversion_unit;
      else                         axisinc = 10  * conversion_unit;
    %end;

/*Round the min's value to match the increment; if the number is */
/* rounded up so that it becomes larger than the lowest data value, */
/* decrease the min by one increment. */
  axislow = round(low,axisinc);
  if axislow > low then axislow = axislow - axisinc;
/* Round the max; if the number is rounded down, */
/* increase the max by one increment. */
  axishigh = round(high, axisinc);
  if axishigh < high then axishigh = axishigh + axisinc;
/* put the values into the global macro variables */
  call symput("&returned_min",compress(put(axislow, best.)));
  call symput("&returned_max",compress(put(axishigh, best.)));
  call symput("&returned_increment",compress(put(axisinc, best.)));
run;
%mend set_axis_minmaxincrement;

SAS Proc GCHART equivalent of Proc GPLOT UNIFORM

1 Answers