How to count the rows by the value in another column

Question

I have been using JMP 12.0 for some time and find it fantastic. However, there is a problem where I would appreciate some help since I seem to be the only person who uses JMP wheras the rest uses SPSS.

I would like to get som information regarding the following:

C1         C2

Title        Computed order

r1   surgeon     1

r2   surgeon     2

r3   surgeon     3

r4   resident     1

r5   resident     2

r6   intern         1

r7   intern         2

etc.

A friend and collegue of mine uses SPSS and has in that program written the following formula which computes the value in C2 in his SPSS.

Do if Title>LAG(Title).

Compute Order=1.

End if.

Do if Title=LAG(Title).

Compute Order=LAG(Order)+1.

End if.

Execute.

How can I write a corresponding formula in JMP?

Very greatful if someone could explain how.

LarsBirger

Kevin_Anderson · Accepted Answer

Hi, Lars!

Here's a script that I think does what you're asking on the Big Class dataset in the Sample Data:

Names Default To Here( 1 );

BC_dt = Open( "$SAMPLE_DATA/Big Class.jmp" );

BCSorted_dt = BC_dt << Sort( By( :age, :sex, :height, :weight ), output table name("Big Class Sorted"));

BCSorted_dt << New Column( "Computed Order",
	Numeric,
	Ordinal,
	Formula( If(row()==1 | :age!=Lag(:age, 1), 1, Lag(:Computed Order, 1)+1) ),
	EvalFormula);

If all you need is the formula, you can copy the Formula() argument in the script and paste it in the formula editor...just remember to sort the data appropriately first.

Good luck!

XanGregg · Accepted Answer

If you're using JMP 13, the Col Rank() function support tie breaking by row number, so you get the effect you're looking for.

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

dt << New Column( "Rank within age",
	Formula( Col Rank( 1, :age, <<Tie("row")) ));

The first argument is the value to rank, which is usually another column, but in this case the ranked value doesn't matter so I used the constant, 1.