topic How to delete observations with same hour in SAS Programming

How to delete observations with same hour

Marian_2127 — Wed, 12 Jun 2013 14:15:37 GMT

Hello everybody,

I'm just trying to delete some obs - the ones that are created at the same date and hour.

There is a column "creation_time" from which I extrat the creation_hour by using

attrib

creation_hour length = 8 format = hour.

;

creation_hour = timepart(creation_time);

(That doesn't work as I expected it to because '11:47:32' becomes '12' but that is not really important for my task.)

But there is an actual problem: Even if '11:47:32' and '11:48:01' both lead to '12' the two obs still remain, if I try to delete one of them by using

proc sort nodupkey: by creation_hour;

That means the column creation_hour actually still contains the whole information, with minutes and seconds, doesn't it?

How can I change this? I really only want the column to contain the numbers from 1-12 - is that possible? How?

Or is there a better way to do the job?

Thanks for any ideas

Re: How to delete observations with same hour

ballardw — Wed, 12 Jun 2013 14:51:15 GMT

Creation_hour= hour(creation_time);

Should return just the hour.

Re: How to delete observations with same hour

Marian_2127 — Thu, 13 Jun 2013 07:02:36 GMT

Thanks for the idea but unfortunately, this doesn't return the hour. Now creation_hour equals zero for all obs - without any errors or anything in the log.

Re: How to delete observations with same hour

andreas_lds — Thu, 13 Jun 2013 10:59:23 GMT

Why do you apply the hour-format to creation_hour? I think the problems are a result of applying the format.

Re: How to delete observations with same hour

Marian_2127 — Thu, 13 Jun 2013 11:09:05 GMT

Ah, I see.

I used the format 'hour' because with

"creation_hour = timepart(creation_time); "

without this format, the column creation_hour contained numbers from zero to 86.399 (counting the seconds).

But with the correction
"creation_hour = hour(creation_time); "

the format is indeed not necessary.

Thanks a lot to both of you!