如何在 SAS 中制作数据透视表

Question

我有下表

+--------+----------+-----------+
| ID     | var_name | var_value |
+--------+----------+-----------+
| 153879 | age      | 35        |
+--------+----------+-----------+
| 153879 | gender   | Male      |
+--------+----------+-----------+
| 153879 | income   | 1000      |
+--------+----------+-----------+
| 13527  | age      | 18        |
+--------+----------+-----------+
| 13527  | gender   | Male      |
+--------+----------+-----------+
| 13527  | income   | 20        |
+--------+----------+-----------+
| 14416  | age      | 40        |
+--------+----------+-----------+
| 14416  | gender   | Female    |
+--------+----------+-----------+
| 14416  | income   | 500       |
+--------+----------+-----------+

如何制作数据透视表，使结果看起来像这样

+--------+-----+--------+--------+
| ID     | age | gender | income |
+--------+-----+--------+--------+
| 153879 | 35  | Male   | 1000   |
+--------+-----+--------+--------+
| 13527  | 18  | Male   | 20     |
+--------+-----+--------+--------+
| 14416  | 40  | Female | 500    |
+--------+-----+--------+--------+

我尝试通过以这种方式反复左加入表格来做到这一点：

data table;
    input id $ var_name $ var_value $;
datalines;
153879 age 35
153879 gender Male
153879 income 1000
13527  age 18
13527  gender Male
13527  income 20
14416  age 40
14416  gender Female
14416  income 500
run;


Proc SQL;
    Create table have as
    Select a.ID
          ,b.var_value as age
          ,c.var_value as gender
          ,d.var_value as income
    From (select distinct id from table) as a
    Left Join (select * from table where var_name = 'age') as b
        On a.id = b.id
    Left Join (select * from table where var_name = 'gender') as c
        On a.id = c.id
    Left Join (select * from table where var_name = 'income') as d
        On a.id = d.id;
Quit;

使用此代码我得到了所需的结果，但是，由于实际表要大几倍，并且包含 50 多个变量，我想知道是否有任何其他方法可以做到这一点，更有效和更短。

Answer 1

用PROC TRANSPOSE 转置您的表格。您首先需要按id 对其进行排序或索引，如果id 和var_name 有任何重复值，则预先聚合这些值。

proc sql noprint;
    create table have_aggregated as 
        select id
             , var_name
             , sum(var_value) as var_value
        from have
        group by id, var_name
        order by id, var_name
    ;
quit;

proc transpose data=have_aggregated
               out=want(drop=_NAME_);
    by id;
    id var_name;
    var var_value;
run;