Assuming day values less than 10 have leading zeroes, something like this should work:
data list list /numdate (f10).
begin data
01012006
02012006
3012006
end data.
string #temp (a8).
compute #temp=string(numdate,N8).
compute #month=number(substr(#temp,1,2), f2).
compute #day=number(substr(#temp, 3, 2), f2).
compute #year=number(substr(#temp, 5, 4), f4).
compute date=date.mdy(#month, #day, #year).
formats date (adate10).
list.
It can be done in fewer lines of code, I just find this easier to interpret and debug.
#temp is a string containing the numeric date value. N8 formats insures that 7-digit values (ones with no leading zero for month) will have a leading zero in the string value.
Then we take the various parts of #temp and create numeric variables that contain the first two digits (#month), the second two digits (#day), and the last four digits (#year).
Finally, we use the date.mdy function to create a real date from the three parts -- but make sure to assign a date format or the value will display as a very large integer (number of second since October 14, 1582).
-----Original Message-----
From: SPSSX(r) Discussion [mailto:
[hidden email]] On Behalf Of Mark Palmberg
Sent: Friday, August 24, 2007 10:47 AM
To:
[hidden email]
Subject: Date transformation
I have a numeric field from a database import that's 10 characters and
in the format 11051991 (for November 5, 1991). But there are also
values like 3151985 (no leading 0) for March 15, 1985. When I try to
apply a date field format as mm/dd/yyyy, all the dates end up sometime
in the 1500s. How do I make these dates useable? This is in SPSS 15.
Thanks.
Locked
