CoStat Without the GUI
(Or, how to do data manipulation and statistics from the command line,
batch files, shell scripts, pipes, Java programs, Perl, Python, Rexx, Tcl, etc.)

You can bypass the graphical front end of CoStat in order to do data manipulate and statistics from the command line, batch files, shell scripts, pipes, Java programs, Perl, Python, Rexx, Tcl, etc.

When you run CoStat, you are really running a graphical front end to a large number of Java classes. You can also get access to all of these classes via a Java class called CoData (which comes with CoStat). There are two ways to use CoData:

• CoData is a text-mode program which can run a CoData macro file (an actual file or a stream from a pipe). CoData's macro language is a subset of Java. This approach is useful if you want to write another program (in any language) to generate a CoData macro file on the fly (for example, on a server), pipe that into CoData, and pipe the results into some other program.
• CoData is a Java class, so you can write another Java source code file which extends CoData, thereby gaining access to all of the procedures in CoData. This approach is useful if you want write a Java program that uses the services that CoData provides (basically, all of the data manipulation and statistical analysis procedures in CoStat). To use this approach, you can write your program in Java and use a Java compiler (Java Developer Kits are freely available from Sun and other companies). Or, you can write your program in another language that can access Java procedures (for example, Perl, Python, Rexx, or Tcl).

One big advantage of programming with CoStat (as opposed to other statistical programs that support programming, like SAS) is that CoStat has procedures to extract the individual results from a statistics procedure, not just the printout of the results. As a simple example: if you do a linear regression, you can get the resulting equation in the form of a string ("4.213 + 3.475*x") or you can get the intercept (4.213), the slope (3.475), and the R^2 value separately. This makes CoStat ideal for use as a statistics engine for your programs.