Scripts to launch a platform should generally specify the columns to analyze. If you submit a script that launches a platform without specifying columns and roles, you get the dialog box for launching the platform. After you choose columns and click OK, you get the analysis that you specified by script. In other words, JMP remembers and obeys any other messages in your script after getting the column assignments that it needs.
Note: The Column() argument for the Distribution platform is not the same as the Column() function. It is similar to the X() or Y() arguments in the Fit Y by X platform.
Throughout this manual, a col placeholder represents any data table column reference, and a nomCol, ordCol, or contCol placeholder suggest that a nominal, ordinal, or continuous column, respectively, would be most appropriate. In many cases, columns of other modeling types would also be accepted. JMP returns an error if you try to cast a column into a role that is strictly not allowed.
Platform scripts that include a Where statement treat column references differently. You must use Eval() on the column reference variables to return the column name. The other option is to use only column names.
This example defines column references for the weight and height columns. The Bivariate platform message is passed to the data table, and the column references are specified as Y and X roles. Notice the Eval() statements that get the column names for Ycol and Xcol.
Y( Eval( Ycol ) ), // get column names for Ycol and Xcol
X( Eval( Xcol ) ),
The command Action, if sent to a platform, simply evaluates the expression, whatever it is. You can use this command to chain invocations to platforms where you want to use the platform launch window to ask the user to choose columns and then continue the script after the platform is launched.
In the following example, the script first asks you to assign columns for each of the four platforms in turn, and then print Done to the log. The script is effectively stopped four times, each time to prompt for columns for a platform launch. At each step, the user fills in a launch window. Four reports are open at the completion of the script.
Platform launches have an invisible option, which suppresses the showing of the window. Using this option on a Fit Model script suppresses both the model launch window and the results window.
The following example extracts the F-Ratio from a bivariate report. Make sure Big Class.jmp is open when running this script.
The invisible option also works on the Tables menu operations, suppressing the creation of the window (rather than just hiding it).
You can specify the title (shown in the title bar of a platform’s report) by adding the title command to the launch request. For example, the following replaces the standard bivariate report’s title with a user-specified one.