为什么ERR：“已定义为字符型和数值型”

1. %do j = 1 %to 15;
   
2.    proc model noprint data = one;  endogenous a;  exogenous r f VA e;
   
3.           e = a*probnorm((log(a/f) + (r+va*va/2))/VA) -
   
4.           f*exp(-r)*probnorm((log(a/f) + (r-va*va/2))/VA);  solve a/out=two;
   
5.    data two;   set two;   num = _n_;   keep a num;
   
6.    data one;   set one;   num = _n_;   drop a;  
   
7.    data two;   merge one two;   by num;  l1p = lag1(permno);   l1a = lag1(a);
   
8.    data two;   set two; if l1p = permno then ra = log(a/l1a);
   
9.    proc means noprint data = two;   var ra;   by permno;   output out = bob;
   
10.    data bar;   set bob; if _stat_ = 'MEAN';  mu = 252*ra;   keep permno mu;
    
11.    data bob;   set bob; if _stat_ = 'STD'; va1 = sqrt(252)*RA;
    
12.                if va1 < 0.01 then va1 = 0.01; keep permno va1;
    
13.    data one;   merge two bob bar; by permno; vdif = va1 - va;
    
14.                if abs(vdif) < 0.001 and vdif ne . then conv = 1;
    
15.    data fin;   set one; if conv = 1;   assetvol = va1; proc sort; by permno descending date;  
    
16.    data fin;   set fin; if permno ne lag1(permno);   curdat = 100*&yyy + &mmm;  iter = &j;
    
17.    data conv;   merge conv fin;   by permno;  drop va ra l1p l1a conv cdt num;
    
18.    data one;   set one; if conv ne 1;   va = va1;   drop va1;
    
19. %end;

复制代码

变量permno是股票的代码。运行到“data fin;   set fin; if permno ne lag1(permno);   curdat = 100*&yyy + &mmm;  iter = &j;” 时会提示：ERROR: 变量 permno 已定义为字符型和数值型。这是什么原因呢？恳请高人解答啊！

原因： 两个数据集参加运算时，相同变量有不同的类形。

/* run the following code and look at the results */
/* to see if there is any difference in the last column */

proc sql;
   select memname, name, type
   from dictionary.columns
   where libname='WORK' and name='permno';
quit;

检查你数据集two bob bar中的permno是不是即有数值型又有字符型

这个错误应该不是“data fin;   set fin; if permno ne lag1(permno);   curdat = 100*&yyy + &mmm;  iter = &j;” 这一步产生的。

Agree with them.
Just guessing: if proc means automatically picks up the format of permno, is it possible for its output giving a different type of permno with the original one?  Of course, it is possible only if permno is numeric in data two that actually originates from data one.
jingju

谢谢各位的答案，我先试试看