Hi, Linlin That's awful! I feel so sorry for you; I've had similar situations over the years, and I know how frustrating it is. Hang in there, it will get solved, and you'll be back in business. Dumb question; can you use SQL*Plus on the troublesome machine to connect with Oracle and retrieve data> Good luck!   Tom ... View more

Yeah, Art! Our fifth column is already preparing the ground for you... http://www.youtube.com/watch?v=BrhA0sEkuaM ... View more

WOW! You REALLY blew up SAS! They're going to cancel your license!! Seriously, assuming it's SAS/Access for Oracle that's choking, could you maybe use ODBC as a workaround until SAS gets it fixed? Good luck; I've never seen one like this in SAS/Access. Be sure to let us know what happened. Best,   Tom ... View more

You mention feeding a data warehouse, which to my view implies large data volumes. My experience with using messaging interfaces to try to pull large data volumes is that they are not designed for this, and are VERY slow. I suggest ensuring that you can meet the required volumes with your approach. Would another alternative be to meet your analysis needs from whatever analysis resources Guidewire supplies? Let them do the heavy lifting? Tom ... View more

Way to go! Just a warning, the "brain hurts" condition is pretty much permanent one you dive deep into SAS. The product is phenomenal, can do pretty much anything, and usually six different ways. Enjoy! One quick correction to your note. Enterprise Guide does create user defined formats, just not the type that use picture specifications.   Tom ... View more

This looks like one of the functions that they haven't yet added to Enterprise Guide. (You can only create "look-up" style formats). There is a custom task available to extend EG for this, but that's probably a bit beyond you at this point. My suggestion is follow Art's advice using SAS code. To do this: -   In Enterprise Guide, select File | New | Program This will open a code editor window. You can just copy and paste the three lines from Art's post right into the window. Make sure you take the line that says "run;". That's really important in SAS. -  Select Run (beside the green arrow) This will execute your code, and in this case create a custom format named MyMSdt. You can then use that anywhere in Enterprise Guide. One important note is that since this code doesn't specify where to store the format, it will be put into your WORK library, which means it will be deleted at the end of your Enterprise Guide session. So you'll have to recreate it next time. If you've learned about permanent libraries, you can store the format in one by changing your "proc format" line to something like "proc format lib=sasuser;" where this will store it in my permanent SASUSER library. Good luck!   Tom ... View more

I'd suggest posting that question to the Statistical Procedures category. My feeling is that you're on very thin statistical ice trying to use missing data to calculate a median, but this discussion forum has people with much stronger statistical chops than me. Tom ... View more

Hi, Karun You can very quickly use some rules of thumb for the data volumes of your original data: You have 32 variables, and a SAS numeric variable is 8 bytes, so if they were all numeric each observation would need 8*32, or 256 bytes. You can see that your observation length is 592 bytes, so there are some character variables in there that are longer than 8. Using your 592 byte observation length, and your 52,829 observations, SAS needs somewhere around 31 MB to store the data (there are always overheads and inefficiencies, so these numbers are never exact). This is perfectly reasonable, and won't cause you any problems. As you noted, when you convert it to XML the file size balloons to 147 MB. Ballardw's comment is exactly correct; XML is a very flexible medium, but the price of that is that it required enormous amounts of storage. I'm surprised that your XML dataset isn't even bigger. Native Excel is much more efficient in storage terms, so your dataset drops again to 8MB (even smaller than in SAS). The price you pay for this efficiency is that Excel files can't be used in as many different places as an XML file. Finally, it's very likely that the size of your SAS dataset could be reduced, maybe significantly. This would involve analyzing your character variables, and reducing the length to that of the longest value, and possibly reducing the length of your numeric variables, based on the characteristics of the contents. But it's a lot of work, and really not worth it for only 52,000 observations. Hope this answers your questions. Tom ... View more

Art is correct, and there are a few more options. From the Describe tab, you can use either Summary Statistics or Summary Tables, depending on the type of results you need (using either the Wizard or just the plain old task on either is a matter of preference...I find the Wizard better for new users, as it has more handholding, but if you already know how to get there, you'll get your request done faster with the non-wizard task). If you're saving results to a dataset, in Summary Statistics you can untick "Show statistics" and you won't get report output. Summary Tables doesn't have the equivalent capability. To be complete, you can also get medians from List Report and Distribution Analysis, but they're somewhat specialized. I agree with Art that your best bet is Summary Statistics. Tom ... View more