分组计数、均值

假设我的数据有如下两列：
var1 var2
A     10
B      11
A      20
C      30
B       15
D      14
B      23
.....

我现在想对var1进行分组计数、并计算每一组的均值。即要生成两个新变量var3和var4，var3用来存放每组对应的频数，var4用来存放每组对应的均值。

请问用data步程序如何做？

以前好像解决过类似的问题，但长时间没接触SAS，又忘记了。

proc summary data = yourdata nway;
    class var1;
    var var2;
    output out = summary(drop = _type_ index = (var1) rename = (_freq_ = var3)) mean = var4;
run;

data result;
    set yourdata;
    set summary key = var1/unique;
    _error_ = 0; if _iorc_ = 0;
run;

in data step looks like not explicit. It is much clear for what it does in 2nd floor.
Of note: if VAR1 contains missing, you have to work extra lines on it.
JingJu

1. data _null_;
   
2.         call symputx('nobs',nobs);
   
3.         if 0 then set have nobs =nobs;
   
4.         call symputx('LenVar1',vlength(var1));
   
5.         stop;
   
6.         run;
   
7. data have2; drop i;
   
8.         array c[&nobs.] $&LenVar1. _temporary_; array n[&nobs.,3] _temporary_;
   
9.         if _n_ =1 then do p =1 to nobs;
   
10.                 set have point=p nobs=nobs;        
    
11.                 do i =1 to dim(c);
    
12.                         if (c[i] =' ' or var1 =c[i]) then do;
    
13.                                 c[i] =var1;
    
14.                                 n[p,3] =i; n[i,1] ++(var2 ^=.);n[i,2] ++var2;
    
15.                                 leave;
    
16.                                 end;
    
17.                         end;
    
18.                 end;
    
19.         set have end=eof2;
    
20.         var3 =n[n[_n_,3],1];
    
21.         var4 =n[n[_n_,3],2]/n[n[_n_,3],1];
    
22.         run;

复制代码

jingju11 发表于 2013-8-14 00:39
in data step looks like not explicit. It is much clear for what it does in 2nd floor.
Of note: if V ...

jingju11, thanks for your answer, but I suddenly find a really easy way to solve this problem, for your codes, I don't understand somewhere, but I will learn it and make it clear, thanks again.

yuerqieqie 发表于 2013-8-13 20:19
proc summary data = yourdata nway;
    class var1;
    var var2;

thanks for your answer.

干嘛非用data步，用sql很方便
proc sql;
create table temp as
select var1,var2,sum(var2),avg(var2)
from yourdata
group by var1

看看这个是不是你想要的？
data a;
input type$ x@@;
cards;
a 3
b 9
a 20
b 15
a 7
b 12
;
run;
proc sort data=a out=b;
by type;
run;
data c(keep=type freq mean);
set b;
by type;
retain freq sum;
if first.type then do;
freq=0;
sum=0;
end;
freq+1;
sum+x;
if last.type;
mean=sum/freq;
run;

你好，我也遇到这种列计数问题，请问您想到的简便方法是如何用data步求var3的？
在线等。

tjnkswordsman 发表于 2013-11-2 11:35
你好，我也遇到这种列计数问题，请问您想到的简便方法是如何用data步求var3的？
在线等。

请问您解决了吗