如何删除SAS数据集Dairy1中变量名的前缀下划线？

It looks like your original code had a couple of key issues that prevented it from working correctly. Let's break down the problems and fix them step by step.

Why Your Original Code Failed #

• Incorrect Filtering: Your WHERE clause '240' = scan(name, 2, '_') only targets the single variable _240, not all variables like _241, _242, etc.
• Invalid Rename Syntax: Using scan(name, 1, '_') on a variable name starting with _ returns an empty string (since the first character is the delimiter), leading to broken rename statements like _240=.
• Case Sensitivity: SAS stores dataset names in uppercase in the dictionary.columns table, so memname = 'Dairy1' won't match your dataset (it’s stored as DAIRY1).

Corrected Solution #

First, generate a valid list of rename statements using PROC SQL, then apply the renames with either a data step or PROC DATASETS (more efficient for just renaming variables).

Step 1: Generate the Rename List

proc sql noprint;
  select cats(name, '=', substr(name, 2))
  into :rename_list separated by ' '
  from dictionary.columns
  where libname = 'WORK'          /* Adjust if your dataset is in another library */
    and memname = 'DAIRY1'        /* Must be uppercase */
    and name like '_%';           /* Select all variables starting with underscore */
quit;

%put &rename_list.; /* Verify output (e.g., _240=240 _241=241) */

Step 2: Apply the Renames

Option 1: Rename variables in-place (no new dataset created):

proc datasets lib=WORK nolist;
  modify DAIRY1;
  rename &rename_list.;
quit;

Option 2: Create a new dataset with renamed variables:

data want;
  set WORK.DAIRY1;
  rename &rename_list.;
run;

Alternative: Rename Without Macro Variables #

If you prefer avoiding macro variables, you can use PROC DATASETS with a direct RENAME statement, but the macro approach is more flexible for dynamic variable lists.

内容的提问来源于stack exchange，提问作者Juan Felipe M Rocha