Hi Alisa, Glad you solved your problem! to convert character to numeric, you need to use input: %let a=20; data _null_;   b=input("&a",2.);   put b; run; ... View more

Hi Tom, Thank you very much!!! - Linlin ... View more

how about: data population_a; input city_code rank_100; cards; 100 1 100 2 100 4 100 8 101 3 101 6 101 9 102 21 102 26 102 30 ; proc sql;   create table temp as     select city_code,rank_100,                    case                                     when city_code=100 then                    case                                              when rank_100 in (1,2,3) then   '1_3'                                                    when rank_100 in (4,5,6) then   '4_6'                                                    when rank_100 in (7,8,9) then '7-9'                                                    else 'no_rank' end                                          when city_code=101 then                    case                                              when rank_100 in (1,2,3) then  '1_3'                                                    when rank_100 in (4,5,6,7,8) then  '4_8'                                                    when rank_100 in (13,14,15) then '13-15'                                                    else 'no_rank' end                                          else                    case                                              when rank_100 in (20,21,22) then  '20_22'                                                    when rank_100 in (23,24,25,26) then  '23_26'                                                    when rank_100 in (27,28,29,30) then '27_30'                                                    else 'no_rank' end                                          end as Rank_rng format=$7.            from population_a;            quit; proc print;run;                     Obs     code    rank_100    Rank_rng                          1     100         1       1_3                          2     100         2       1_3                          3     100         4       4_6                          4     100         8       7-9                          5     101         3       1_3                          6     101         6       4_8                          7     101         9       no_rank                          8     102        21       20_22                          9     102        26       23_26                         10     102        30       27_30 Linlin ... View more

/* I add a observation to your original data:  SLOW INCRE   23DEC2007    10051001 */ /* borrowed Haikuo's dataset */ data have; input txt &$11. DPFDATE  :date9.  SUBJID :$8.; format dpfdate date9.; cards; SLOW INCRE   13DEC2006    10051001 ROUTINE DO   14DEC2006    10051001 ROUTINE DO   15DEC2006    10051001 ROUTINE DO   16DEC2006    10051001 ROUTINE DO   17DEC2006    10051001 ROUTINE DO   18DEC2006    10051001 ROUTINE DO   19DEC2006    10051001 ROUTINE DO   20DEC2006    10051001 ROUTINE DO   21DEC2006    10051001 ROUTINE DO  22DEC2006    10051001 ROUTINE DO  23DEC2006    10051001 ROUTINE DO  24DEC2006    10051001 ROUTINE DO  25DEC2006    10051001 ROUTINE DO  26DEC2006    10051001 ROUTINE DO  27DEC2006    10051001 ROUTINE DO  28DEC2006    10051001 ROUTINE DO  29DEC2006    10051001 ROUTINE DO  30DEC2006    10051001 ROUTINE DO  31DEC2006    10051001 SLOW INCRE   23DEC2007    10051001 ; data temp; set have; n=_n_; proc sort; by txt n; data want;   set temp; by txt; if (first.txt or last.txt); run; proc print;run; Obs       txt          DPFDATE     SUBJID      n 1     ROUTINE DO    14DEC2006    10051001     2 2     ROUTINE DO    31DEC2006    10051001    19 3     SLOW INCRE    13DEC2006    10051001     1 4     SLOW INCRE    23DEC2007    10051001    20 Linlin ... View more

Hi mspak, Have you tried the code? I don't understand why you think it is complicated to use the code. what you need to do is just replacing the libname and dataset with your own, then all the unwanted variables are dropped. there are other ways from the link http://communities.sas.com/message/119802#119802 Good luck! ... View more

Hi mspak, Below is Ksharp's code to drop variables with all missing values. You can modify it to drop variables with certain number of missing values. Note:  the libname and dataset name have to be  CAPITAL LETTERS in " where libname='WORK' and memname='HAVE';" replace "&nobs" with  "your- number" in "  if _x{i} eq &nobs then _list=catx(' ',_list,vname(_x{i}));". you need to increase the length of " _list" in "length _list $ 4000;". data have; infile cards missover; input   Name  $   DOB  :$10.    (Add1  Add2 Addr3) (:$); cards;      John     10/10/80          data           data      Alan      11/11/81          data      Paul      10/10/79          data ; run; proc sql noprint; select catx(' ','nmiss(',name,') as',name) into : list separated by ','   from dictionary.columns    where libname='WORK' and memname='HAVE'; select count(*) into : nobs from have; create table temp as   select &list from have; quit; data _null_; set temp; length _list $ 4000; array _x{*} _numeric_; do i=1 to dim(_x);   if _x{i} eq &nobs then _list=catx(' ',_list,vname(_x{i})); end; call symputx('drop',_list); run; data want; set have(drop=&drop); run; Linlin ... View more