Arrange in rows by group

[#1]

Hi,

Here is the problem I am trying to solve:

6 Y continuous variables (Y1,Y2,..Y6), same continuous X variable (1) and sample ID field (SID).

I am doing a BiVariate fit grouping by SID. I would like all 6 plots to be in a row and with each SID group.

SID=1

Y1,Y2,Y3,Y4,Y5,Y6,

SID=2

Y1,Y2,Y3,Y4,Y5,Y6,

......

How would I script this?  The Y's and X are know and fixed while the number of sample IDs will vary.

Andy

[#2 · Accepted solution]

Here is an example script using a methodlogy I have used many times, that I think produces the results you want.

Names Default To Here( 1 );
dt = Open( "$SAMPLE_DATA/semiconductor capability.jmp" );

// Setup data to look somewhat like your data
dt << delete columns( 1, 2, 3 );
Column( dt, 1 ) << set name( "SID" );
Column( dt, 2 ) << set name( "X" );
Column( dt, 3 ) << set name( "Y1" );
Column( dt, 4 ) << set name( "Y2" );
Column( dt, 5 ) << set name( "Y3" );
Column( dt, 6 ) << set name( "Y4" );
Column( dt, 7 ) << set name( "Y5" );
Column( dt, 8 ) << set name( "Y6" );
dt << delete columns( Index( 9, 129 ) );

// Determine the number of SIDs
Summarize( dt, SIDgrps = by( SID ) );

// Create the output window
nw = New Window( "Report", MainVLB = V List Box() );

// Create a list to hold the data table names 
dtList = {};

// Setup what to do when the display window is closed
nw << on close( For( i = 1, i <= N Items( SIDgrps ), i++, Close( dtList[i], nosave ) ) );

// Loop through the SIDs and generate the required reports
For( i = 1, i <= N Items( SIDgrps ), i++,
	dt << select where( :SID == Num( SIDgrps[i] ) );
	dtSub = dt << subset( selected rows( 1 ), selected columns( 0 ), invisible );
	Insert Into( dtList, dtSub );
	dtSub << set name( "SID=" || SIDgrps[i] );
	ob = Outline Box( "SID=" || SIDgrps[i],
		H List Box(
			dtSub = Bivariate( Y( :Y1 ), X( :X ) ),
			dtSub = Bivariate( Y( :Y2 ), X( :X ) ),
			dtSub = Bivariate( Y( :Y3 ), X( :X ) ),
			dtSub = Bivariate( Y( :Y4 ), X( :X ) ),
			dtSub = Bivariate( Y( :Y5 ), X( :X ) ),
			dtSub = Bivariate( Y( :Y6 ), X( :X ) ),

		)
	);
	// Add the current set of reports to the main display
	MainVLB << append( ob );
);

[#3 · Accepted solution]

From your description, it seems like all you have to do is use SID in the By role.

Bivariate(
	Y( :Y1, :Y2, :Y3, :Y4, :Y5, :Y6 ),
	X( :X ),
	By( :SID )
);

[#4]

The code from txnelson does the job,  Though in his example I needed comment out the code line

//dt << delete columns( Index( 9, 129 ) ); for it to run.

I must admit, I don't fully understand it and I stumbled a bit with this line

dt << select where( :SID == Num( SIDgrps[i] ) ); since my ID is actually character.  I removed the Num() to get it to work.

Mark's solution only does the by SID grouping but does not group the plots in rows.

Andy

[#5]

This shows one way to control the layout but the resulting array of plots is the transpose of what you want.

New Window( "In a Row",
	Line Up Box( N Col( 2 ),
		Bivariate(
			Y( :age, :weight, :oxy, :runtime ),
			X( :runpulse ),
			By( :sex )
		)
	)
);

[#6 · Accepted solution]

I think this way might be simpler and give you the layout you want:

New Window( "Fitness - Bivariate",
	V List Box(
		Fit Group(
			Bivariate( Y( :Age ), X( :MaxPulse ) ),
			Bivariate( Y( :Weight ), X( :MaxPulse ) ),
			Bivariate( Y( :Oxy ), X( :MaxPulse ) ),
			Bivariate( Y( :Runtime ), X( :MaxPulse ) ),
			Bivariate( Y( :RunPulse ), X( :MaxPulse ) ),
			Bivariate( Y( :RstPulse ), X( :MaxPulse ) ),
			<<{Arrange in Rows( 2 )},
			By( :Sex )
		),
		Fit Group(
			Bivariate( Y( :Age ), X( :MaxPulse ) ),
			Bivariate( Y( :Weight ), X( :MaxPulse ) ),
			Bivariate( Y( :Oxy ), X( :MaxPulse ) ),
			Bivariate( Y( :Runtime ), X( :MaxPulse ) ),
			Bivariate( Y( :RunPulse ), X( :MaxPulse ) ),
			Bivariate( Y( :RstPulse ), X( :MaxPulse ) ),
			<<{Arrange in Rows( 2 )},
			By( :Sex )
		)
	)
);