Mark ,

I said it shouldnt be done, not that it should never be done. Your point 1 and 2 are valid I will concede those points, some times you just have to do what the boss and/or client wants, been there done that. On the otherhand (geez I sound like Daivd) it still doesnt mean it should be done that way, it is just the programmer has no choice in the decision. If it is just because a person thinks it will make their life easier or they dont know any other way then a redesign of the code or rethinking of their process is in order. Which is the point I was making albiet probably not nearly as clear as I should have.

Now the call execute thingy might or might not work. Depends on the number of discrete values which will have to be muliplied by no less than three ( number of lines of code to subset the data one time ). This could put too much on the stack and SAS will poop out. Or one could wrap the three lines of code into a macro and use call execute delaying the macro execution until it is popped off of the stack ( one line of code for each descrete value) if it is more than SAS can handle SAS will poop out again. Or one could simply write the code to a file and %include that back into the program.

All I can say is I am sure glad as hell I dont have to mess around with 200 datasets for one project, the nightmares alone would keep me up late at night.

Toby Dunn

Quickly, bring me a beaker of wine, so that I may wet my mind and say something clever. Aristophanes

Wise people, even though all laws were abolished, would still lead the same life. Aristophanes

You should not decide until you have heard what both have to say. Aristophanes

From: "Terjeson, Mark" <Mterjeson@RUSSELL.COM> Reply-To: "Terjeson, Mark" <Mterjeson@RUSSELL.COM> To: SAS-L@LISTSERV.UGA.EDU Subject: Re: Creating multiple datasets from one dataset based on value of a single variable Date: Tue, 14 Nov 2006 10:09:48 -0800

Hi Toby,

We can't always know the background and objectives. Normally I don't comment on different viewpoints of why someone wants to do something, and many of us have frequently shared this facet numerous times regarding keeping things normalized instead of breaking into many separate files. However, there are three issues that may prompt such a desire. The first two are 1) the boss said to, 2) the recipient of the data requested that format. The third, 3) since Mike mentioned millions of rows, it just may actually prove more effective and efficient in their environment to process and deal with smaller chunks. Just as is, or they may have applications that work with the data selectively and smaller chunks is just plain faster. Who knows. So we'll arm them with the best design information and arm them with code to break it up, and they will have to apply the best decisions. Your idea of shouldn't be done usually applies and once I had only a total of a few hundred records and even after sharing recommendations, the boss said "that's what the customer wants" and it was my job to say "Okay". :o)

Hope this is helpful.

Mark Terjeson Senior Programmer Analyst, IM&R Russell Investment Group

Russell Global Leaders in Multi-Manager Investing

-----Original Message----- From: SAS(r) Discussion [mailto:SAS-L@LISTSERV.UGA.EDU] On Behalf Of toby dunn Sent: Tuesday, November 14, 2006 9:48 AM To: SAS-L@LISTSERV.UGA.EDU Subject: Re: Creating multiple datasets from one dataset based on value of a single variable

Mike ,

The reason you are having such a hard time finding a post along this lines is because it shouldnt be done ( by the way there are a few out there) . By breaking your data into so many little parts all you will accomplish is to east up space on your hard drive and make you life as well as others infinitly more complicated by making yoru code more complicated to write, understand, debug, and modify. Now I have no clue as to why you think you need to do this but I have a sneaking suspicion that if you write back with a brief description of what you are trying to accomplish by breaking these data up we can and will happily show you a different way that keeps your data together and will make you life easier.

Toby Dunn

Quickly, bring me a beaker of wine, so that I may wet my mind and say something clever. Aristophanes

Wise people, even though all laws were abolished, would still lead the same life. Aristophanes

You should not decide until you have heard what both have to say. Aristophanes

From: MikeInWV <mikeslover@GMAIL.COM> Reply-To: MikeInWV <mikeslover@GMAIL.COM> To: SAS-L@LISTSERV.UGA.EDU Subject: Creating multiple datasets from one dataset based on value of a single variable Date: Tue, 14 Nov 2006 09:41:26 -0800

I've looked in the archives but can't seem to find a solution. If there is one I apologize!

Here is an example of what I am looking at:

************************************************* data all; input client $1. val 2.; cards; a1 b2 c3 a4 a5 a6 a7 a8 a9 b10 b11 c12 ; run; **************************************************** Obviously this creates a single dataset called "all" and contains 2 variables with 12 observations--for 3 different clients (A,B,C). Now, in my case I have about 7 million observations and about 200 different clients. I want to create 200 different datasets, each of which is based on the value of "client". And of course I want each dataset to equal to the value of the client. So, if I continued with the initial dataset above I want something that gives me the equivalent of this:

data A B C; set all; if client='A' then output A; else if client='B' then output B; else if client= 'C' then output C; run;

I DONT want to create 200 different else if statements especially since I know that the large dataset will change over time so it needs to be flexible. I'm sure there is a macro trick out there but I just can't think of it.

Thanks

Mike

_________________________________________________________________ All-in-one security and maintenance for your PC. Get a free 90-day trial! http://clk.atdmt.com/MSN/go/msnnkwlo0050000002msn/direct/01/?href=http:/ /www.windowsonecare.com/?sc_cid=msn_hotmail

_________________________________________________________________ Add a Yahoo! contact to Windows Live Messenger for a chance to win a free trip! http://www.imagine-windowslive.com/minisites/yahoo/default.aspx?locale=en-us&hmtagline