Instead of putting your libname statement in the "submit when server connected" option, put it explicitly into a code window, and submit it to your server. Your log should then show an error message, which you can post to give us some indications of the problem. Tom ... View more

One important point; although you don't add code in a SQL Join to sort the data, in any non-trivial application the data wil be sorted behind the scenes to enable the merge. I've seen cases where people think they save resources by eliminating one or more sorts, and this is absolutely not the case. Tom ... View more

In the situation I had, it was in the ODBC connection. You'll need to check with your DBMS documentation to find the right parameter. Tom ... View more

I can confirm SASKiwi's comment. I researched this for a client once, and you can't create two tables from one SQL select statement. In this case, your DATA step is the best solution, unless there's another reason for wanting to use SQL? Tom ... View more

If it is a small quantity of records (<10,000): 1. Define your SQL Server database and schema as a SAS library, using SAS/Access (I assume you've already done this). 2. Use either data step or proc sql: libname sqllib ...; /* data step */ data sqllib.newdata; set work.olddata; run; /* proc sql */ proc sql; create table sqllib.newdata as select * from work.olddata; quit; I don't have access to SAS in my environment, so there may be some syntax errors. But is should be close. Of course, SAS being SAS, there are many other ways to do it. The 10,000 record limit is somewhat arbitrary, based on my experience. If the above option is too slow, there is an option to use the SQL Server bulk loader. I've never used in in SQL server, but in another environment the bulk loader was 100 to 1000 times faster. Tom ... View more

If your connection is set up to pre-read very few records, bumping that number might help. I got a huge increase once going from pre-reading 10 records to 500. Frequently the pre-read is set low because DBMSs tend to be used for transaction processing, not bulk retrievals. Tom ... View more

Almost! Instead of using your new variable as a classification variable, use it as an analysis variable. Put it beside your column variable, and make that statistic a sum instead of n (because it's always 0 or 1, a sum represents a count.) Here's a screen cap of my report designer (it looks funny because I have EG 4.1 at home, sigh!) However, this will only get you the count. I didn't realize you needed the percentage for that variable, and I've got to admit I can't think of a way to get it. I'll keep thinking about it, and hopefully someone else will have a good idea. One definite way to do it is with PROC REPORT, which is a VERY sophisticated reporting program in SAS, but I think that would be very challenging for someone new to SAS. Tom ... View more

1. Use a format. This is a very convenient feature of SAS, and of EG, and it's really worth learning. You can read about it in the SAS help. There is an EG task to create one, or you can use SAS code like the following (assuming that your variable is numeric; slightly different syntax if it's character). PROC FORMAT LIB=WORK; VALUE METFMT 1 = "Not Met" 2 = "Met" 3 = "High Met"; RUN; and then apply it to your variable at creation time or in your summary tables task. 2. Are you using the summary tables task to create the percentages? If so, I don't understand why the percentage formats don't work. Maybe somebody else can shed some light on this. 3. This is going to be tricky. It would be easy if you wanted the total of all three, but just getting 2 and 3 is a challenge. I suggest you use the query builder to create a dataset based on your dataset, but with an additional column (let's call it ALLHIGH) that you create with an advanced expression. In the advanced expression, use the CASE "function" to create a variable that is 1 if your source variable is 2 or 3, 0 otherwise. Then, in your summary table task, drag ALLHIGH to the RIGHT of your current column variables, not above or below, and set the calculation to a sum. I think that'll work. ... View more

If you have access to it, take a look at PROC EXPAND. It's designed for operations like this, but it's hard to tell if it meets your specific requirement. Tom ... View more

Just.This week, I noticed a change in adifferent Access engine. I think there may be some changes from 9.1 to 9.2. ... View more

Try using the CSV destinations. I don't think there is a performance penalty for using them with large datasets. Tom ... View more