If you want to make this interactive from a web interface, with SAS underneath, then you may need some additional SAS tools.
I expect you can prototype a bunch of stuff with EG, but I also expect that you will need to ultimately program directly in SAS.
As an intro to SAS programming:
SAS is a data centric language, not process and control. The focus is on "this is my data, this is what I want to do with the data" not "I have a box, this is what I want the box to do".
This is why the base programming block stucture is a "Data" step, and opening and closing files are generally implicit operations, not explicit.
There is a new user around here -- Joshua -- who's picked himself up by his bootstraps. I would recommend searching for his postings to find out which books he's used/read to get himself started.
Chris@SAS is around here as well, and he is one of the favored authors.
I started programming in PL/C back in 1977, and have touched/programmed/reviewed a large number of languages, including (but not limited to, and in no particular order) Pascal, C, COBOL, SNOBOL, ForTran, Ada, Forth, Tutor, Basic, 4th Dimension, FoxBase, awk, Korne shell, Perl, VisualBasic, C++, VBscript and Java. My favorite languages include SAS, C and Java, each of equal standing to the other (right tool for the job kind of thing). I said many years ago "SAS is the greatest data processing language on the planet".
Is it necessary to perform the work through SAS? You have a web server and a database, so perhaps what is lacking might be an application server that encapsulates the business logic and is able to pull the necessary information from the database. I'd try JBoss with Mysql leveraging on innodb storage engine. Thats just my humble opinion as the resident noob.
Josh has a point. You could use SAS in the background to build the regression equations (say run every night) and then store the coefficients to a table. The web ap could then just do the arithmetic to get the expected values for the specific parameters set by the customer. If your customers are sufficiently sophisticated, you might also store the covariance matrix so you could compute error bars for the estimates.
Thank you for the replies. It's very helpful.
I really thought that I won't be needing any middleware between the webapplication and the SAS program. Do I really need that or isn't there any other solutions to get on with this?
I wanted to make the calculation directly when there is a new user trying to make a valuation of a house. I really coan't understand why I should run batches during night?