如何在SAS Enterprise Guide（SAS企业指南）中从数据集中选择行，并将特定条件应用于每个子集

我有下表：

COMPANY_NAME | GROUP | COUNTRY | STATUS  
COM1         |   1   |    DE   | DELETED   
COM2         |   1   |    DE   | REMAINING  
COM3         |   1   |    UK   | DELETED  
COM4         |   2   |    ES   | DELETED  
COM5         |   2   |    FR   | DELETED  
COM6         |   3   |    RO   | DELETED  
COM7         |   3   |    BG   | DELETED  
COM8         |   3   |    ES   | REMAINING  
COM9         |   3   |    ES   | DELETED 

我需要得到：

COMPANY_NAME | GROUP | COUNTRY | STATUS  
COM3         |   1   |    UK   | DELETED  
COM4         |   2   |    ES   | DELETED  
COM5         |   2   |    FR   | DELETED  
COM6         |   3   |    RO   | DELETED  
COM7         |   3   |    BG   | DELETED

---

因此，我需要状态已删除的所有条目，并且在每个组中，没有任何公司名称的状态与已删除状态保持在同一国家/地区。我可以使用PROC-SQL或数据步骤

到目前为止，我尝试的是：

PROC SQL;
CREATE TABLE WORK.OUTPUT AS
SELECT *
FROM WORK.INPUT
WHERE STATUS = 'DELETED' AND COUNTRY NOT IN (SELECT COUNTRY FROM WORK.INPUT WHERE STATUS = 'REMAINING');
QUIT;

但这显然将所有其他国家排除在所有集团之外

我还尝试了一个数据步骤：

DATA WORK.OUTPUT;
SET WORK.INPUT;
BY GROUP COUNTRY;

IF NOT (STATUS = 'DELETED' AND COUNTRY NOT IN (COUNTRY WHERE STATUS = 'REMAINING')) THEN DELETE; 

RUN;

但是语法不正确，因为我不知道正确的书写方法。

试试这个：

proc sql;
select * from your_table
where status = 'deleted' and 
      catx("_",country,group) not in 
         (select catx("_",country,group) from your_table where status='remaining');
quit;  

输出：

company_name | group | country | status
com3         |   1   |    UK   | deleted
com4         |   2   |    ES   | deleted
com5         |   2   |    FR   | deleted
com6         |   3   |    RO   | deleted
com7         |   3   |    BG   | deleted

试试这个：

proc sql;
select * from your_table
where status = 'deleted' and 
      catx("_",country,group) not in 
         (select catx("_",country,group) from your_table where status='remaining');
quit;  

输出：

company_name | group | country | status
com3         |   1   |    UK   | deleted
com4         |   2   |    ES   | deleted
com5         |   2   |    FR   | deleted
com6         |   3   |    RO   | deleted
com7         |   3   |    BG   | deleted

---

你的解决方案表明你的思路是正确的

一个数据步骤解决方案是：

data want(drop = remain_list);
   length remain_list $ 20;

   do until(last.group);
      set have;
      by group;

      if status = 'REMAINING' and not find(remain_list, country) then
         remain_list = catx(' ', remain_list, country);
   end;

   do until(last.group);
      set have;
      by group;

      if status = 'DELETED' and not find(remain_list, strip(country)) then
         output;
   end;
run;

---

你的解决方案表明你的思路是正确的

一个数据步骤解决方案是：

data want(drop = remain_list);
   length remain_list $ 20;

   do until(last.group);
      set have;
      by group;

      if status = 'REMAINING' and not find(remain_list, country) then
         remain_list = catx(' ', remain_list, country);
   end;

   do until(last.group);
      set have;
      by group;

      if status = 'DELETED' and not find(remain_list, strip(country)) then
         output;
   end;
run;

---

目标中的第一条记录是否应该是

COM3

，而不是

COM1

？请点击DomPazz。我改了。谢谢。“我可以使用PROC SQL或数据步骤”您尝试了哪一个？请共享代码和所有错误消息。^若要添加，StackOverflow不是一种代码编写服务。如果您不知道从何处开始，请将状态代码更改为可以排序的数字代码，以便最小值或最大值以及只有一个是您想要的。然后就很容易解决了。如果目标中的第一条记录是

COM3

，而不是

COM1

，请点击DomPazz。我改了。谢谢。“我可以使用PROC SQL或数据步骤”您尝试了哪一个？请共享代码和所有错误消息。^若要添加，StackOverflow不是一种代码编写服务。如果您不知道从何处开始，请将状态代码更改为可以排序的数字代码，以便最小值或最大值以及只有一个是您想要的。然后就很容易解决了。@reeza为什么我们需要将状态代码转换为数字？很乐意帮助：）@reeza为什么我们需要将状态代码转换为数字？很乐意帮助：）谢谢。我选择了procsql解决方案，因为它的代码更少。但是很高兴看到数据步骤的执行。谢谢。我选择了procsql解决方案，因为它的代码更少。但是看到数据步骤的执行也很棒。