Question

我无法让proc报告完成我想做的事情。

我有一张表，其中包含州，项目，计数，州和百分比。有总结线给出按州和总计的总数。我的问题是这些摘要线总结了总计水平的州总数。像这样：

CODE：

proc report data=dataset nowd ;
columns state item count pct_state percent;

define state /order 'State';
define item / 'Status';
define count / '#';
define pct_state / '% of State';
define percent / '% of Total';

break after state/ol summarize;
compute after state;
  item=catt(state,' Total');
  state = '';
  line @1 ' ';
endcomp;
rbreak after /ol summarize;
compute after;
  involved = 'Grand Total';
endcomp;
run;

制作一个这样的表：

State  Item   #  %state  %total
AL      A     2   40.0%   20.0%
        B     3   60.0%   30.0%
     AL Total 5  100.0%   50.0%

MN      A     1   20.0%   10.0%
        B     1   20.0%   10.0%
        C     3   60.0%   30.0%
     MN Total 5  100.0%   50.0%

Grand Total   10 200.0%  100.0%

正如您所看到的，它将状态％total报告为200％，这是一个荒谬的数字。我宁愿让它根本不总结州的价值。我知道sas网站警告在使用汇总行的表格上使用日期，因为SAS将它们解释为数字变量并因此总结它们......但它没有提供一个好的解决方案。我真的不明白为什么BREAK和RBREAK语句没有允许你指定的“VAR”选项......但现在我需要一个解决方法。

我想出的是创建一个新变量并将百分比存储为文本，以便在摘要中无法计算，但这是一种非常倒退的方式。

data dataset; set dataset;
  state_txt = trim(left(put(pct_state,percent10.1)));
run;

proc report data=dataset nowd ;
columns state item count state_txt percent;

define state /order 'State';
define item / 'Status';
define count / '#';
define state_txt / right '% of State';
define percent / '% of Total';

break after state/ol summarize;
compute after state;
  item=catt(state,' Total');
  state = '';
  line @1 ' ';
endcomp;
rbreak after /ol summarize;
compute after;
  involved = 'Grand Total';
endcomp;
run;

这消除了所有的摘要（因为它是一个字符变量）但是当我应该能够说rbreak after /summarize var=count percent;这样的事情并且完成它时，它似乎只是一种可怕的做事方式。有没有更好的方法呢？另外，我不介意它是否将每州的水平概括为100％......虽然它不是优先考虑的重点，但远不如说它在底部不是200％（或者在完整的情况下）美国表，5000％）。

示例数据：

data dataset;
 length state item $50;
 infile datalines delimiter=','; 
 input state item $ count percent pct_state;
 datalines;  
  AL,A,8,0.0047,1.0000
  DC,A,1,0.0006,0.5000
  DC,B,1,0.0006,0.5000
  FL,A,18,0.0107,0.7500
  FL,B,2,0.0012,0.0833
  FL,C,4,0.0024,0.1667
  LA,A,434,0.2576,0.8314
  LA,B,69,0.0409,0.1322
  LA,C,19,0.0113,0.0364
  MI,A,1,0.0006,1.0000
  MS,A,4,0.0024,0.8000
  MS,B,1,0.0006,0.2000
  OK,A,2,0.0012,1.0000
  PA,A,1,0.0006,1.0000
  TX,A,943,0.5596,0.8435
  TX,B,132,0.0783,0.1181
  TX,C,43,0.0255,0.0385
  VA,A,1,0.0006,1.0000
  WI,B,1,0.0006,1.0000
  ;

Answer 1

我认为在你的COMPUTE AFTER中使用一些逻辑可以解决这个问题。试试这个（我稍微改变了数据，如果这不代表你的数据，请告诉我）：

（在out =语句中左侧，这可能会有所帮助）

data dataset;
 length state item $50;
 infile datalines delimiter=','; 
 input state item $ count percent pct_state;
 format percent pct_state percent10.1;
 datalines;  
  AL,A,8,0.8,1.0000
  DC,A,1,0.1,0.5000
  DC,B,1,0.1,0.5000
 ;

proc report data=dataset nowd out=work.report;
 columns state item count pct_state percent;

  define state /order 'State';
  define item / 'Status';
  define count / '#';
  define pct_state / '% of State';
  define percent / '% of Total';

  break after state/ol summarize;
  compute after state;
   item=catt(state,' Total');
   state = '';
   line @1 ' ';
  endcomp;

  rbreak after /ol summarize;
  compute after;
   State = 'Grand Total';
   if pct_state.sum>1 then pct_state.sum=1;
  endcomp;

 run;

与PROC REPORT和汇总线搏斗

1 个答案: