顶一顶，，虽然不懂。但是期待大神帮我们版主解决问题。

1. 先按照var6 的四个级别 level 分开
    楼主是要按 var6 的四个level进行分组计算， 还是要分成四个数据集？
如果是后者的话， 可以这样.

data Level_A  Level_B  Level_C  Level_D;
    set test;
        if var6 = "Level A" then output Level_A;
   else if var6 = "Level B" then output Level_B;
   else if var6 = "Level C" then output Level_C;
   else if var6 = "Level D" then output Level_D;
run;

如果SAS的内存不够，可以先分成几个小数据集。然后再分组，再合并。

data subset_1;
    set test (firstobs=1 obs=100);
run;
data subset_2;
    set test(firstobs=101 obs=200);
run;
data subset_3;
    set test(firstobs=201);
run;



2. 按 var6 分组计算的 percent
proc sql;
    select var1, var2, var3, var4, var5, var6,
           sum(case when var5=0 then 1 end)  as count_V5_0,
           sum(case when  var5 ne 0 then 1 end) as count_v5_,
           count(var5) as nv5,
           calculated  count_v5_0 / calculated nv5 as percent_zero,
           calculated  count_v5_ / calculated nv5 as percent_nonzero
    from test
    group by var6;
quit;