Check for the value of the ENCODING option in your EG workspoace server session. Submit the following code:   proc options option=encoding; run; You want this to be  24 proc options option=encoding; run; SAS (r) Proprietary Software Release 9.4 TS1M4 ENCODING=UTF-8 Specifies the default character-set encoding for the SAS session. NOTE: PROCEDURE OPTIONS used (Total process time): real time 0.00 seconds cpu time 0.00 seconds but likely you get    24 proc options option=encoding; run; SAS (r) Proprietary Software Release 9.4 TS1M4 ENCODING=LATIN1 Specifies the default character-set encoding for the SAS session. NOTE: PROCEDURE OPTIONS used (Total process time): Your SAS administrator will have to make the appropriate changes to the server configuration. This is not a trivial change if this is a server based installation as there is impact on existing data and potentially multiple users will be involved. If you are running a local server you will be the only one but still existing data sets with Latin encoding will have to be migrated or performance will be hurt. So in any case it is not a simple toggle.   Hope this helps, -- Jan. ... View more

Please ask a question. The more specific the better. Stating a desire rarely fulfills that desire.   Regards, -- Jan ... View more

Your info is very lomited but it does ring a bell. Please elaborate if my response misses the point.   Your program tries to create an output file in the default location (current working directory) of the SASApp server. Users are generally not allowed to create files in that location, nor is it recommended.   IF this program is run through the sas data step batch server than configure the log and print locations such that they end up in a more suitable place. This can either be somewhere under the user's home directory (I'm not enthousiastic about that idea) or a specific location dedicated to this purpose. Talk to your SAS administrator about this configuration aspect.   The output location can be configured via SASMC. It's in the properties of DATA Step Batch server.   Hope this helps, -- Jan. ... View more

The datastep is not the tool of choice to calculate these univariate statistics. Your SQL is quite appropriate.   The better non-SQL way would be to use proc means. This procedure can calculate your stats (and then some) and output them to a dataset with ease. I will give you the code I do not want to withhold an opportunity to read the docs and learn.   Hope this helps, -- Jan. ... View more

Not to my knowledge. Never seen it. But you may decide to use the "complete" or "noprompt" option to your ODBC libname statement where you have the entire odbc.ini file inside your SAS code. Any variable content could then be dealt with via macro variables.   See Usage Note 52777 for details.   An example would be: libname mydata odbc complete="driver=SQL Server; user=&SYSUSERID; pwd=password; database=SQL-database; Server=server-name" schema=schema-name; Will be a challenge to stick in the password without spilling that into source code and log files.   Hope this helps, -- Jan.     ... View more

Date variables are not stored in a formatted way. They are stored numerically as the number of days since 01JAN1960. Only when presenting them in a report or a table viewer a format is used. Conversion is not needed when you want to reconsider the presentation format of your variable. In your case the format MONYY5. would the appropriate one. MONYY7. would also print the century.   Formats are a fundamental concept in SAS. Have a look at the doc if you want to read up on how they are used in SAS.    Hope this helps, -- Jan. ... View more

@JacobH wrote: Presumably because the macros are not defined for each observation, but for the dataset as a whole?      Hi Jacob,   It's how the SAS macro processor works. SAS code is first handled by that macro processor before the data step runs. By the time the macro processor is has done its job all macro references are removed, either by executing statements (%let ...) or by resolving variable references (&macro_start_...). Be careful to distinguish what happens when. This is a common mistake and I suggest you read up on macro processing in general. The resolution of your actual challenge, creating variable names based on data values, will be much more evident after that.    On a side note I confess that variables that contain data values are my main pet peeve. They tend to make reporting and analysis very hard.   Hope this helps, -- Jan ... View more

I dare you with this denormalized model. Remember there are brand_1 and brand_2 variables that need to be combined. I look forward to your mean MEANS. And don't make it non-trivial!   Regards, -- Jan. ... View more