BookmarkSubscribe
Choose Language Hide Translation Bar
Craige_Hales
Staff (Retired)

Re: Log showing error, but manual execution does not flag error

pretty sure you want this

dt<<Fit Group

and not this

dt<<Bivariate

FitGroup is a lot like a platform and needs to know what table to use to make the by groups. FitGroup then runs the contained bivariate platforms, once for each by group. The by groups are actually subset tables, and using dt<<Bivariate makes the platforms run on the entire table, not the subset that FitGroup prepared.

I have not tried this with multiple open data tables, so there may be still another issue.

 

 

Craige
0 Kudos
Highlighted
Hegedus
Community Trekker

Re: Log showing error, but manual execution does not flag error

Attached is a data table with a table script.

 

Generally the plots look ok, the conern is the log is showing lots of errors.  I do not like releasing code to production that is producing errors.

 

Andy

0 Kudos
Craige_Hales
Staff (Retired)

Re: Log showing error, but manual execution does not flag error

Appears to be a bug in JMP 12, fixed in 13. It looks like JMP's FitGroup attempts to evaluate the by statement, in addition to actually using it to make by groups. You can work around the problem like this:

dt=open("C:\Program Files\SAS\JMPPRO\12\Samples\Data/big class.jmp");
by=0; // work around JMP 12 issue
New Window( "Big Class - Bivariate",
  V List Box(
    dt<<Fit Group(
      Bivariate( Y( :height ), X( :age ) ),
      Bivariate( Y( :weight ), X( :age ) ),
      <<{Arrange in Rows( 2 )},
      By( :name( "sex" ) )
    )
  )
);

Now JMP can evaluate the variable "by" (and ignore the zero result) without producing messages in the log. 

Edit: make it clear this problem is in FitGroup. 

Craige