0 
rsubmit;
proc sql;
create table wpffasdekpr2 as
select polnum, ordnnr, brkd, fsnr, fvpkd, mnr, mlopnr,coverfrom, coverto, nationalkey, exposuretype, COActivityCode, sum(COSumInsured) as SumIns, sum(AnnualBaseAmt) as SumAmt from wpffasdekpr
group by 
polnum, ordnnr, brkd, fsnr, fvpkd, mnr, mlopnr,coverfrom, coverto, nationalkey, exposuretype, COActivityCode;
quit;
endrsubmit;

此操作的表如下所示：

在此处输入图像描述

我希望前 4 行通过这个语句压缩成一个，并且两个最右边的变量被求和。但是，该声明没有任何效果。输出表看起来像输入表。两个问题： 1. 为什么会这样？在通常的sql中，我很肯定它会起作用。2. 我怎样才能达到预期的效果？

日志：

508  rsubmit;
NOTE: Remote submit to SESKUX01 commencing.
1391  proc sql;
1392  create table wpffasdekpr2 as
1393  select polnum, ordnnr, brkd, fsnr, fvpkd, mnr, mlopnr,coverfrom, coverto, nationalkey, exposuretype, COActivityCode, sum(COSumInsured) as COsumInsured2, sum(AnnualBaseAmt) as AnnualBaseAmt2
1393! from wpffasdekpr
1394  group by
1395  polnum, ordnnr, brkd, fsnr, fvpkd, mnr, mlopnr,coverfrom, coverto, nationalkey, exposuretype, COActivityCode;
NOTE: Compressing data set WORK.WPFFASDEKPR2 increased size by 5.17 percent.
      Compressed is 29614 pages; un-compressed would require 28159 pages.
NOTE: Table WORK.WPFFASDEKPR2 created, with 28778049 rows and 14 columns.

1396  quit;
NOTE: PROCEDURE SQL used (Total process time):
      real time           1:54.66
      cpu time            1:41.68


NOTE: Remote submit to SESKUX01 complete.
4

1 回答 1

2

CoverFrom 和 CoverTo 是时间还是日期值？如果时间，它们可能不相同。可能值得检查是否有任何格式应用于源数据集中的其他变量，因为 group by 操作将处理基础值（而不是格式化/显示的结果）。

如果做不到这一点，我建议重新执行您的查询，一次使用一组。此外，使用以下语法（编号分组依据）可以避免意外错误：

 proc sql;
 create table wpffasdekpr2 as
    select polnum, ordnnr, brkd, fsnr, fvpkd, mnr, mlopnr,coverfrom, coverto
       ,nationalkey, exposuretype, COActivityCode
       ,sum(COSumInsured) as SumIns, sum(AnnualBaseAmt) as SumAmt 
    from wpffasdekpr
    group by 1,2,3,4,5,6,7,8,9,10,11,12;
于 2014-02-13T15:58:53.687 回答