The data a store an index information and data b is actually the data.

You want to get the data out of b according to the index in a. That b is symmetric is not relevant here I suppose.

Here is a solution.

1) load data into a temporary array (matrix).
2) grab the data accoding to v1, v2;

Here is a sample program and I define b matrix in 10*10. and so the index range with be in [1,10]. Mydata is your v3.

Hope this helps.


data a;

do i=1 to 5;
   v1=ceil(10*ranuni(101));
   v2=ceil(10*ranuni(101));
   output;
end;
drop i;
run;

data b;
array x(10);
do k=1 to 10;
do i=1 to 10;
     x(i)=ceil(100*ranuni(101));
end;
output;
end;
drop i k;
run;

title 'a=index of Row and Column of b';
proc print data=a; run;
title 'b is a 10*10 square matrix';
proc print data=b; run;

data need;
   array datmt (10,10) _temporary_;
   do until (end);
      set b end=end;
          array x(10);
          n+1;
          do i = 1 to dim(x);
             datmt (n,i)=x(i);
          end;
   end;
   do until (end2);
      set a end=end2;
          mydata=datmt (v1,v2);
          output;
   end;
   keep v1 v2 mydata;
run;

proc print; run;

2# bobguy

太谢谢了。翻了手册和帮助，不知这么理解思路对吗？

data need;
   array datmt (10,10) _temporary_;
   do until (end);
      set b end=end;
          array x(10);
          n+1;
          do i = 1 to dim(x);
             datmt (n,i)=x(i);
          end;
   end;                                                 /*这段是把数据集b转换成二维数组，要通过一维数组过度*/
   do until (end2);
      set a end=end2;
          mydata=datmt (v1,v2);
          output;                                     /*用二维数组向新变量赋值。只是不明白的是为什么不能直接mydata=datmt(v1,v2)，而要用循环语句？*/
   end;
   keep v1 v2 mydata;
run

我没有编程基础，刚刚开始学SAS编程。希望多多指教。

data a;
do v1=1 to 10;
  do v2=1 to 10;
   output;
  end;
end;
run;
data b(drop=i j);
array t x1-x10;
do i=1 to 10;
  do j=1 to 10;
   t[j]=round(rannor(12345)*2.5+10,1);
  end;
  output;
end;
run;
data a(keep=v1 v2 v3);
set a;
temp=v1;
array tmp x1-x10;
set b point=temp;
v3=tmp[v2];
run;

1) 不知这么理解思路对吗？
You should be able to design cases to check it is Right or Wrong. It should be a problem.

2)/*这段是把数据集b转换成二维数组，要通过一维数组过度*/
Think about that a SAS table is two dimentional. The first set loop just dumps the table into a 2-dimentional temporary array so that it could be referred anytime in PGM.

The following programs are equivalent
*1);
data b2;
   set b;
run;
*2);
data b3;
  do until(end);
   set b end=end;
   output;
  end;
run;

Type 1)  is usually you see in SAS book. Actually SAS do a lot of thing in a set statement. For example, SAS internal loop, where it puts a output statement implicitly, etc. The most important thing is SAS accesses its data set sequencially(without point option in set).

Type 2)  just writes out more explicitly.There are certain avantages. For example, in this example, load data into a temporary array for look-up is much simpler that conventional ways.

I see a quite few of OLD programmers like that way because it is more like a COBOL style. And some people believe it is faster/more efficient than type 1.  

3) /*用二维数组向新变量赋值。只是不明白的是为什么不能直接mydata=datmt(v1,v2)，而要用循环语句？*/

I think the answer is in 2) if you understand type 2)

Beyond SAS manuals, Rick Aster wrote a SAS/BASE book "Professional SAS Programming Secrets" many years ago based on version SAS6.6, this is the only book I would recommend. Unfortunately it is quite old. He may update it ...

enjoy!

HTH.

5# bobguy


That is a model program. So admirable!

To Joskow:

3) /*用二维数组向新变量赋值。只是不明白的是为什么不能直接mydata=datmt(v1,v2)，而要用循环语句？*/

I may have more words on this question.
The loop you have mentioned, with the explicit output statement, forcibly read and output data A with variable ‘mydata’ till its end. If without the explicit loop, after reading the first record in A, (SAS) control is passed to the top of the program, where the end of data B is detected at the time(because it is already at its end), SAS terminates the program thus only one record is created, as you might have seen.


Only my personal understanding.