Fit x by y on subset of data vs other subset of data

[#1]

Hi, this is not a JSL question like most of my others are.

I have some data, and I'd like to fit x by y on it, one half of the data vs the other half of the data. I realize a caveman way of doing this is cut pasting the second half of the data into a new column. But I'd really rather not adjust the table like that. Any solutions? I was advised to use the Split function under Tables so I could toy around with that, but is that all there is to it? Anything more specific?

Thank you hugely in advance, and thank you so much for all the help you've been providing to this date. You are an amazing and amazingly patient community.

Best,

Mike

[#2 · Accepted solution]

If I'm reading your question correctly, it sounds like you have a single column of data that has both your X values and Y values in it.

Do you have a column that identifies whether a row is an X or Y value? Do you have another column (or combination of columns) that identifies which XY pair that row belongs to?

For example, in the table below Value has both X and Y values. Measure identifies whether a row is a height or weight value, and name, age, sex together identify a unique entity for that height or weight.

If this is the kind of data you have then yes, Tables->Split is exactly what you need.

You would complete the Split dialog with Measure in the Split By role, Value in the Split Columns role, and name, age, sex in the Group role. 

name	age	sex	Measure	Value
KATIE	12	F	height	59
KATIE	12	F	weight	95
LOUISE	12	F	height	61
LOUISE	12	F	weight	123
JANE	12	F	height	55
JANE	12	F	weight	74
JACLYN	12	F	height	66
JACLYN	12	F	weight	145
LILLIE	12	F	height	52
LILLIE	12	F	weight	64
TIM	12	M	height	60
TIM	12	M	weight	84
JAMES	12	M	height	61
JAMES	12	M	weight	128
ROBERT	12	M	height	51
ROBERT	12	M	weight	79
BARBARA	13	F	height	60
BARBARA	13	F	weight	112
ALICE	13	F	height	61
ALICE	13	F	weight	107
SUSAN	13	F	height	56
SUSAN	13	F	weight	67
JOHN	13	M	height	65
JOHN	13	M	weight	98
JOE	13	M	height	63
JOE	13	M	weight	105

[#3]

Is there a basic way to do this in JSL? Split()?

[#4]

Actually I am following the syntax in the Scripting Index and results are not as expected. The split in Split(Split()) excludes that column, instead of isolating it, and it doesn't seem it shows the split by column. It's just pretty messy and I don't know that it's working properly.

[#5]

After completing the Tables->Split by hand one time, you can find a Source script in the resulting data table. That will have the script for doing the split. No need to write it by hand. You can copy/paste it into your script.

This is true for most of JSL. Do it by hand then get the script for it from JMP.

Let JMP write your script! 

[#6 · Accepted solution]

Hi @mostarr,

Ya, split is basically it. @Jeff_Perkinson as given a way to do it if you know what to split by. If there is only a single column in your data, then like copy and paste will be the quickest.

[#7]

Thanks for the responses.

I found that a specific type of Split worked perfectly. Now I am wondering if there is a way to save the script automatically? Or do I have to rewrite it manually in JSL?

Mike

[#8 · Accepted solution]

Another option besides splitting your datatable is creating a new column to randomly group your rows (or could systematically group them by row counts) into different groupings and use that grouping as an overlay such as in this example:

// Open Data Table: Big Class.jmp
// → Data Table( "Big Class" )
Open( "$SAMPLE_DATA/Big Class.jmp" );

// New column: Group
Data Table( "Big Class" ) << New Column( "Group",
	Character,
	"Nominal",
	Formula( If( Random Integer( 2 ) == 1, "Group A", "Group B" ) )
);

// Report snapshot: Big Class - Graph Builder
Graph Builder(
	Size( 534, 456 ),
	Show Control Panel( 0 ),
	Variables( X( :height ), Y( :weight ), Overlay( :Group ) ),
	Elements(
		Points( X, Y, Legend( 5 ) ),
		Line Of Fit( X, Y, Legend( 7 ), Equation( 1 ) )
	)
);