928905_en-US

liliconcom

grouped Shuffle

In Big Class, students have to write a letter to another student with the same age (student A writing to student A is OK).
Which Colum Formula can I use to add name_2?

The formula should be fast even with mio of rows.

Automation and ScriptingRe: grouped Shuffle

workaround: dummy table - shuffle - join

Names Default to Here(1);
dt1 = Open( "$SAMPLE_DATA/Big Class Families.jmp" );
New Column( "Rank1",	Formula( Col Rank( 1, :age ) ));
New Column( "Rank2",	Formula( Col Rank( random integer (100000), :age ) ));

New Column( "name 2", Character,	"Nominal",	Formula( :name ));
New Column( "age 2", Formula( :age ));

dt2 = dt1 << Subset( All rows, Selected columns only( 0 ) );

dt1 << Join(
	With( dt2 ),
	Match Flag( 0 ),
	Suppress main table formula evaluation( 0 ),
	Select( :name, :age ),
	SelectWith( :name 2, :age 2 ),
	By Matching Columns( :age = :age, :Rank1 = :Rank2 ),
	Drop multiples( 0, 0 ),
	Include Nonmatches( 1, 1 )
);

 

Re: grouped Shuffle

Workaround that doesn't work: Self join
why?

Names Default to Here(1);
dt1 = Open( "$SAMPLE_DATA/Big Class Families.jmp" );
New Column( "Rank1",	Formula( Col Rank( 1, :age ) ));
New Column( "Rank2",	Formula( Col Rank( random integer (100000), :age ) ));

New Column( "name 2", Character,	"Nominal",	set each value( :name ));
New Column( "age 2", set each value ( :age ));

//dt2 = dt1 << Subset( All rows, Selected columns only( 0 ) );

dt1 << Join(
	With( dt1 ),
	Select( :name, :age ),
	SelectWith( :name 2, :age 2 ),
	By Matching Columns( :age = :age 2, :Rank1 = :Rank2 ),
	Drop multiples( 0, 0 ),
	Include Nonmatches( 1, 1 )
);

Re: grouped Shuffle

Combine Col At() and Col Shuffle()

Names Default To Here(1);

dt = open("$SAMPLE_DATA/Big Class.jmp");

new_col = dt << New Column("Shuffle",
	Character,
	Nominal,
	Formula(Col At(:name, Col Shuffle(:age), :age))
);

I didn't verify the results or test the speed

Re: grouped Shuffle

I was stuck with col shuffle():
how can I convert the grouped index to a row index?

Ah, sure, use the grouped index as input for col at(idx, byGroup) -  perfect match!
Nice example for the new functionality.