Hi Patrick, If my data looks like: data nordered; input c b a; cards; 10 555 8762 11 877 9748 12 898 2874 ;run; Then, my data should be changed as: Obs      a      b      c 1     8762    555    10 2     9748    877    11 3     2874    898    12 I don't have any other variable selection criteria. Thanks! ... View more

Thanks! It works! ... View more

Thanks for getting back to the discussion. I've learned a lot from you! ... View more

Thanks, guys! You're amazing! ... View more

Thanks, Linlin. Your code solves the problem. ... View more

data list; input fname $ id $; cards; 1122 bg 1620 bg 1660 bg 0520 tz 0240 tr 0280 tr 0560 tr ;run; proc contents data=list;run;                              Alphabetic List of Variables and Attributes                                     #    Variable    Type    Len                                     1    fname       Char      8                                     2    id          Char      8 filename x pipe 'dir D:\ex\sas_ex\Ksharp\*.dbf /s /b'; data have; length fname $8. id $8.; infile x truncover; input path $100.; fname=scan(path,-2,'.\'); id = scan(path,5,'\'); run; proc contents data=have; run;                              Alphabetic List of Variables and Attributes                                     #    Variable    Type    Len                                     1    fname       Char      8                                     2    id          Char      8                                     3    path        Char    100 proc print data=have; run;                         Obs    fname    id                 path                         1     1122     bg    D:\ex\sas_ex\Ksharp\bg\1122.dbf                         2     1620     bg    D:\ex\sas_ex\Ksharp\bg\1620.dbf                         3     1660     bg    D:\ex\sas_ex\Ksharp\bg\1660.dbf                         4     0240     tr    D:\ex\sas_ex\Ksharp\tr\0240.dbf                         5     0280     tr    D:\ex\sas_ex\Ksharp\tr\0280.dbf                         6     0560     tr    D:\ex\sas_ex\Ksharp\tr\0560.dbf                         7     0520     tz    D:\ex\sas_ex\Ksharp\tz\0520.dbf                         8     1122     tz    D:\ex\sas_ex\Ksharp\tz\1122.dbf proc sql; create table final_selection as select path from have where fname in (select fname from list) and   id in (select id from list); quit; proc print data=final_selection; run;                                Obs                 path                                 1     D:\ex\sas_ex\Ksharp\bg\1122.dbf                                 2     D:\ex\sas_ex\Ksharp\bg\1620.dbf                                 3     D:\ex\sas_ex\Ksharp\bg\1660.dbf                                 4     D:\ex\sas_ex\Ksharp\tr\0240.dbf                                 5     D:\ex\sas_ex\Ksharp\tr\0280.dbf                                 6     D:\ex\sas_ex\Ksharp\tr\0560.dbf                                 7     D:\ex\sas_ex\Ksharp\tz\0520.dbf                                 8     D:\ex\sas_ex\Ksharp\tz\1122.dbf I still see the problem here. From the list, fname='1122' is of 'bg'. In the final_selection data set, look at the obs=8: D:\ex\sas_ex\Ksharp\tz\1122.dbf. The whole point of this project is not to include the obs=8. But I still see the '1122.dbf' again from the 'tz' folder! Am I missing something here? Thanks. ... View more

Thank you for the reply. I already changed it. I check the Ksharp's pipe statement does not collect all the path+dbf information. If I have 0345.dbf both in folder1 and in folder3, the pipe command reads only one of two a.dbf files. What we have to do is to create the path variable, which has data values as follows: ... c:\test\folder1\0345.dbf ... c:\test\folder3\0345.dbf ... But the pipe fails to collect both. *update* And I found that the pipe command does not collect all the dbf files in a single folder. For example, my folder1 has 13 dbf files. But after running the pipe command just for the folder1 (filename x pipe 'dir C:\test\folder1\*.dbf /s /b';), I have only 6 records in the path variable. ... View more

filename x pipe 'dir c:\test\*.txt /s /b'; This statement does not seem to collect all the path+files if files with same name are scattered across the folder1 - folder3. ... View more

The X, Y variables are numeric variables. There are lots of dbf files in many folders. I have to automate this process. Thank you. ... View more

Thanks all. The attached "mydata" is the actual data file. Sorry for the confusion. Anyway, by taking Astounding's advice of INPUT statements, I could first read the U1-U4 variables as character, and then convert them into numeric using the INPUT statements: U1new = input(U1, 9.). ... View more