DatePart

[RickyS]

Did some looking around and did not see anything on this, building an expression in EG 7.1.

Trying to figure out how to create a new MonthYr date value variable from a time date a stamp.

MDY(MONTH(date_variable),1,YEAR(date_variable)

Have used the above code in the past when it is a mm/dd/yy stamp, the time part is throwing me off, how do I build in the DatePart?  I could do it in 2 steps but I'm sure there is a way to make it 1

[PGStats]

Use

datepart(intnx("DTMONTH", date_variable, 0))

[TomKari]

Congratulations! I believe this is a first. You actually solved your problem in your own query!

The DATEPART function: Extracts the date from a SAS datetime value.

Tom

[ballardw]

And I bet I can make a custom format that works directly with a datetime variable so that for many purposes no new variable is needed at all.

But I am not sure what your desired "monthyr" would look like.

[RickyS]

Perhaps I'm missing something

/* Create_MonthYr */

(DATEPART(t1.created_at)) FORMAT=MONYY7. LABEL="Create_MonthYr" AS Create_MonthYr, 

This gives me a JAN2018 format but it still has a SAS value associated with multiple days 01/01/2018, 01/02/2018, 01/03/2018 etc. this makes it more difficult to do some basic reporting like aggregating transaction counts by monthyr across other variables.

Creating a common month year SAS date value will enable this aggregation. I've done this before with MDY function when it is a date, looking for a solution for time date

[ballardw]

---

@RickyS wrote:

Perhaps I'm missing something

/* Create_MonthYr */

(DATEPART(t1.created_at)) FORMAT=MONYY7. LABEL="Create_MonthYr" AS Create_MonthYr, 

This gives me a JAN2018 format but it still has a SAS value associated with multiple days 01/01/2018, 01/02/2018, 01/03/2018 etc. this makes it more difficult to do some basic reporting like aggregating transaction counts by monthyr across other variables.

Creating a common month year SAS date value will enable this aggregation. I've done this before with MDY function when it is a date, looking for a solution for time date

---

I think you need to explain what the difficulty is with your aggregating. Almost every procedure will create groups based on the formatted value of a variable as defined in the proc.

SAS may not supply the exact datetime format you want but proc format with a Picture format will allow a great deal of customization.

data junk;
   do date='03OCT2017:01:23:27'dt to '15JAN2018:18:24:16'dt by 1385;
      value = rand('uniform');
      output;
   end;
run;
proc format library=work;
picture dtmonyr (default=7)
low-high='%b%Y' (datatype=datetime)
;
run;
proc means data=junk  mean max std;
   class date;
   format date dtmonyr.;
   var value ;
run;

[RickyS]

Did it as 2 steps

DATEPART(t1.date)

MDY(month(t1.date),1,year(t1.date))

this converts everything to 1st date of each month