请教SAS 数据处理问题

数据大概如下：

Name   Test1 ..... Test1000
Billy   1              2
.....
....

其实对于所有的 Test Variable, 只有三个值：1,2,3

我想知道，有什么简单方法，可以计算对于每一列（TEST1, TEST2....,TEST1000)的数据中，1出现的百分比，2出现的百分比，3出现的百分比，同时还要把它存储到这个DATA SET的最后一行

对了，数据中有MISSING DATA的，计算百分比的时候，MISSING VALUE不可以计算在内

非常感谢！！

1. *generating a test data set;
   
2. data a;
   
3. array test{1000};
   
4. do j = 1 to 2000;
   
5.   do i = 1 to dim(test);
   
6.    r =ceil (ranuni(0)*12);
   
7.    r2 =ceil (ranuni(0)*11);
   
8.    test{i} = 1+mod(r, 3);
   
9.    if r2 = 11 then test{i} = .;
   
10.    end;
    
11.   output;
    
12.   end;
    
13. drop r:;
    
14. run;
    
15. data b;
    
16.    set a end = Eof;
    
17.    array test{1000}; *load vars;
    
18.    array n_{1000}; *count # of non-missing for each var;
    
19.    array n1_{1000}; *count # of 1s for each var;
    
20.    array n2_{1000}; *count # of 2s for each var;
    
21.    array n3_{1000}; *count # of 3s for each var;
    
22.    array p1_{1000}; *percentage of 1s for each var;
    
23.    array p2_{1000}; *percentage of 2s for each var;
    
24.    array p3_{1000}; *percentage of 3s for each var;
    
25.    do i = 1 to dim(test);
    
26.       if ^missing(test{i}) then
    
27.          do;
    
28.             n_{i}+1;
    
29.             n1_{i}+(test{i} = 1);
    
30.             n2_{i}+(test{i} = 2);
    
31.             n3_{i}+(test{i} = 3);
    
32.             end;
    
33.       end;
    
34. if Eof then
    
35.     do i = 1 to dim(test);
    
36.        p1_{i} = n1_{i} /n_{i};
    
37.        p2_{i}  = n2_{i} /n_{i};
    
38.        p3_{i}  = n3_{i} /n_{i};
    
39.        end;
    
40. drop i j;
    
41. run;

复制代码

If I understand the question correctly, those percentages should converage to 0.33 when randomly generated 1,2, 3 and missing values as what I was doing above;
something wrong with the website, many 's were gone...??

AMAZING!!!!

进来学习了

I really hope to be at ur level of SAS......thank you so much!!!


2# jingju11

xuexile, xiexie!

i have a question, it seems like that i cant use n_ in the do loop, and have to change to be n_{i}, is there any tricky for it?

also, what would be the final data set look like? as following?

test1 ............ test 1000 n_1 ..................n_1000 n1_1......................n1_1000...................


and each one has 2000 obs?

thanks a lot

2# jingju11

rockfido 发表于 2010-3-2 03:12
i have a question, it seems like that i cant use n_ in the do loop, and have to change to be n_{i}, is there any tricky for it?

also, what would be the final data set look like? as following?

test1 ............ test 1000 n_1 ..................n_1000 n1_1......................n1_1000...................


and each one has 2000 obs?

thanks a lot

2# jingju11

Nothing tricky here. Originally I used 中括号，but some problem to show this sign [].
should be like that.

Great programing skill, fantastic

so in that case, all the first 1999 obs are not valid but only the last line is needed? because, I guess for n_, the 2000 obs should be valuing from 1, 2, 3, until 2000?

do u have any idea how could i add those p1_, p2_, p3_ as the last three observations under each variables (instead of being added as 3 new variables....)

thanks a lot!

8# jingju11