Please run Proc Contents on the SAS data set that represents File1. It will have the details that will let us provide specifics on how to manipulate the data. Just because Excel displays a value as a date does not always mean that the SAS variable will because of some vagaries in how many people structure Excel files and the specific method used to bring the data into a SAS data set.
Notice that you saying that you selected a specific date to start with as a reference and that you need dates from that reference does now make much more sense.
The code below shows how to make a data set containing your "dates" as SAS date values and then select using your rule.
As example code the reference date variable is also in the output so you can compare if that is behaving correctly.
data file1;
informat date yymmdd10.;
input date aa;
format date yymmddD10.;
datalines;
1990-02-02 23
1990-02-07 45
1990-02-13 56
1990-02-28 7
1990-03-12 8
1990-03-13 8
1990-03-16 9
1990-03-19 55
1990-03-23 78
1990-04-03 90
1990-04-23 76
1990-05-01 6
1990-05-11 4
1990-05-15 33
1990-05-25 34
1990-05-30 12
;
run;
data file2;;
set file1;
/* Retain will keep the value of a variable
from record to record. Note that to use
a literal date value the ddMONyy in quotes with
D to tell SAS the value should be treated
as a date. None of the other date formats
are acceptable
*/
retain referencedate '07FEB1990'd;
format referencedate yymmddD10.;
if date = referencedate OR date-Referencedate ge 25 then do;
output;
Referencedate = date;
end;
run;
If your proc contents or examination of the data set's variable properties does not show the date as numeric you will need to use an INPUT statement such as
Sasdate = input(date, yymmdd10.); to create a SAS date valued variable and then do all the other calculations using the SASDATE variable.
