Thank you for your help. I got it. Have a nice weekend. HHC ... View more

Thank you, Tom. Yes, it is exactly what I want to do. Still a lot to learn for me. HHC ... View more

Thank you all for different suggestion. Since I try to practice the approach below (which I learn from Tom and other), I really appreciate if you could debug that code. Have a nice weekend. HHC data Number; set time; n=0; i+1; set have; if enter<time and time<exit then do; n=n+1; end; run; ... View more

Hi Tom, Thank you so much for your help. It works nicely. Have a nice day. HHC ... View more

I just get back and check. The file type is: Microsoft Excel Comma Separated Value File. When I open it, it opens in Excel and therefore I see the " time appear on screen is 0:00 for 12:00:00 AM, 1:00 for 1:00:00 AM. I see the full minutes and second (1:00:00 AM) when I click on that cell." To check if it has comma, I right click the file and chose to OPEN with Notepad. Then the data look like that. So look like it depends on the software I use to open, I see different things. 2005.06.28,00:00 2005.06.29,1:00 2005.06.30,5:00 In this case what should I do? Thank you, Tom. HHC ... View more

Thanks, Tom. When I try your code as below, it still doesn't work. All missing on Time column. One thing in csv file is that: the time appear on screen is 0:00 for 12:00:00 AM, 1:00 for 1:00:00 AM. I see the full minutes and second (1:00:00 AM) when I click on that cell. So I don't know if it is something that cause my trouble. HHC data originaldata; infile 'C:\test.csv' delimiter=',';    input date :yymmdd8. time time11.;    format date yymmdd8.;    format time  time8.; run; ... View more

Thank you, I think I got it now. proc sort data=have; by target;run; proc summary data=have; by target; class a1 a2 a3 a4 a5 a6; ways 2; output out=want (drop=_TYPE_); run; /*this step will help you to put all name in a field. you can feel free to delete a1-a6*/ data final; set want; length name1 name2 $5; array _a{*} a:; count=0; do i=1 to dim(_a);   if _a{i}^=. then do; count=count+1;   if count=1 then do;   name1=vname(_a{i});   value1=_a{i};   end;   else if count=2 then do;   name2=vname(_a{i});   value2=_a{i};   end; end; end; run; ... View more

Actually, there are 2 levels of combinations: :smileyinfo: combination of variables and (ii) combination of value of variable. For the first level, I want all combination of a1-a6, which are: a1a2, a1a3,a1a4,a1a5,a1a6; a2a3,a2a4,a2a5,a2a5; a3a4,a3a5,a3a6; a4a5,a4a6; and a5a6. For each of the above variable-combination I want value combination. Take the first pair a1a2 as a example: a1 has value 1 2 3 4 5 8 a2 has value 1 2 3 9 I want to report how many observation with target=1 and target=0 for each value-combination as below: a1=1 and a2=1: 0 observation with target=0 ; 0 observation with target=1 (In fact, for simplicity this report can be ignored since there is no such combination in the data) a1=1 and a2=2:  1 observation with target=0; 0 observation with target=1 …. a1=8 and a2=9: … So I want a FULL combination of both levels (variables and value). HHC ... View more