proc sql语句在数据清洗中的几个运用

摘自《Cody's data cleaning techniques using sas》

 

libname clean "c:/books/clean";                                           /*定义永久性数据库*/

*新建一个样本数据one;
data one;
   input X Y Z;
datalines;
1 2 3
101 202 303
44 55 66
444 555 666
;
title "Values of X from data set ONE where X is greater than 100";
/*有条件的选择观测值*/

proc sql;
   select X
   from one
   where X gt 100;               
quit;

*Program 8-2;
***检查无效的字符型的数值;
title "Checking for Invalid Character Data";
proc sql;
   select Patno,
          Gender,
          DX,
          AE
   from clean.patients
   where Gender not in ('M','F',' ')           or
         notdigit(trim(DX))and not missing(DX) or
         AE not in ('0','1',' ');
quit;

*检查无效的数值型的数值;
title "Checking for out-of-range numeric values";
proc sql;
   select Patno,
          HR,
          SBP,
          DBP
   from clean.patients
   where HR  not between 40 and 100 and not missing(HR)    or
         SBP not between 80 and 200 and not missing(SBP)   or
         DBP not between 60 and 120 and not missing(DBP);
quit;

 

*基于标准差利用简单的算法来检查数值;
title "Data values beyond two standard deviations";
proc sql;
   select Patno,
          SBP
   from clean.patients
   having SBP not between mean(SBP) - 2 * std(SBP) and
      mean(SBP) + 2 * std(SBP)                     and
      SBP is not missing;
quit;

 

*检查缺失值;
options linesize=84;
title "Observations with missing values";
proc sql;
   select *
   from clean.patients
   where Patno   is missing or
         Gender  is missing or
         Visit   is missing or
         HR      is missing or
         SBP     is missing or
         DBP     is missing or
         DX      is missing or
         AE      is missing;
quit;

 

*检查日期;
title "Dates before June 1, 1998 or after October 15, 1999";
proc sql;
   select Patno,
          Visit
   from clean.patients
   where Visit not between '01jun1998'd and '15oct1999'd and
         Visit is not missing;
quit;

 

*检查重复值;
title "Duplicate Patient Numbers";
proc sql;
   select Patno,
          Visit
      from clean.patients
      group by Patno
      having count(Patno) gt 1;
quit;

 

*识别对应多个观察值的变量;
title "Listing of patients who do not have two visits";
proc sql;
   select Patno,
          Visit
      from clean.patients2
      group by Patno
      having count(Patno) ne 2;
quit;

 

*检查两个文件中对应要求的序列号ID;
data one;
   input Patno X Y;
datalines;
1 69 79
2 56 .
3 66 99
5 98 87
12 13 14
;
data two;
   input Patno Z;
datalines;
1 56
3 67
4 88
5 98
13 99
;

*两个文件都不含的ID;
title "Patient numbers not in both files";
proc sql;
   select One.patno as ID_one,
          Two.patno as ID_two
   from one full join two
   on One.patno eq Two.patno
   where One.patno is missing or Two.patno is missing;
quit;