topic Re: Need help writing optimal code for extracting R^2 and Slope from Bivariate Fit in Discussions

Need help writing optimal code for extracting R^2 and Slope from Bivariate Fit

miguello — Fri, 09 Jun 2023 17:00:28 GMT

All,

I'd like to ask the community to nudge me in the right direction.

I need to extract R^2 and Slope form the report and display it on the plot itself like so:

I'm going to do this with parsing Display Trees and then looking into Add Graphics Script to the plot itself.

But, since this task seems to me very standard and routine, I have to ask couple of questions before digging into Display Trees:

1. Is there a more straight forward way of doing that? Like Combined Data Table (which gives slope but does not give R^2) for getting number or like Line Of Fit( X, Y, Legend( 5 ), R²( 1 ), Equation( 1 ) ) in graph builder (customer wants to use Bivariate Fit)

2. Maybe somebody already has a snippet of code ready to use for this task?

Re: Need help writing optimal code for extracting R^2 and Slope from Bivariate Fit

jthi — Wed, 01 Jun 2022 19:49:42 GMT

There are at least three different ways to get those values:

Get them from Bivariate Report with your method of choice
Get them by using Make Into Data Table/Make Combined Data Table and get them from the data tables created
Use matrix function

In this case I would either use option 1 or 2 because you will have to create the Bivariate anyway

Re: Need help writing optimal code for extracting R^2 and Slope from Bivariate Fit

Craige_Hales — Wed, 01 Jun 2022 22:23:43 GMT

You can get really close with Graph Builder

Equation includes the slope.

Re: Need help writing optimal code for extracting R^2 and Slope from Bivariate Fit

txnelson — Thu, 02 Jun 2022 00:21:19 GMT

If you just need the values, and not the graph, you can use the Linear Regression() function, and get the values from there

Names Default To Here( 1 );

// Open Data Table: big class.jmp
// → Data Table( "big class" )
dt = Open( "$SAMPLE_DATA/big class.jmp" );
/*Simple Linear Regression: y = intercept + beta * x + error*/
y = :height << get values;
X = :weight << get values;
{Estimates, Std_Error, Diagnostics} = Linear Regression( y, X, <<printToLog ); 
 
RSquare = Diagnostics["RSquare"];
slope = Estimates[2];

rsquare = 0.502917160762804;
slope = 0.135507698709946;

Re: Need help writing optimal code for extracting R^2 and Slope from Bivariate Fit

miguello — Thu, 02 Jun 2022 06:17:50 GMT

Yes, I scripted that already (as I mentioned in the initial question it's easy by adding Line Of Fit( X, Y, Legend( 5 ), R²( 1 ), Equation( 1 ) ) into the Graph Builder Code), however, customer doesn't like Graph Builder for some reason and insists on using Bivariate Fit.

Re: Need help writing optimal code for extracting R^2 and Slope from Bivariate Fit

miguello — Thu, 02 Jun 2022 06:20:29 GMT

I need graph too, and it has to be Bivariate Fit. Previously the customer was taking a screenshot of the plot and portion of the report below to have Slope and R^2 included (basically image in my initial question without red markings). I'm automating the whole thing and I want to get rid of the report and just write those two values on the plot itself (as illustrated on my screenshot with red text on the plot).

Re: Need help writing optimal code for extracting R^2 and Slope from Bivariate Fit

txnelson — Thu, 02 Jun 2022 07:48:47 GMT

Here is an example that should get you started.

Names Default To Here( 1 );

// Open Data Table: big class.jmp
// → Data Table( "big class" )
dt = Open( "$SAMPLE_DATA/big class.jmp" );

biv = Bivariate( Y( :height ), X( :weight ), Fit Line( {Line Color( {212, 73, 88} )} ) );
Report( biv )[framebox( 1 )] << add graphics script(
	Text Color( {212, 73, 88} );
	xMin = Current Report()[axisbox( 2 )] << get min;
	yMin = Current Report()[axisbox( 1 )] << get min;
	xMax = Current Report()[axisbox( 2 )] << get max;
	yMax = Current Report()[axisbox( 1 )] << get max;
	textList = {};
	Insert Into( textList, xMin + (xMax - xMin) * .25 );
	Insert Into( textList, yMax - (yMax - yMin) * .1 );
	R2 = Current Report()["Summary of Fit"][Number Col Box( 1 )][1];
	R2 = "R^2 = " || Format( R2, "Fixed Dec", 5, 3 );
	Text( Right Justified, textList, R2 );
	textList = {};
	Insert Into( textList, xMin + (xMax - xMin) * .32 );
	Insert Into( textList, yMax - (yMax - yMin) * .15 );
	Slope = Current Report()["Parameter Estimates"][Number Col Box( 1 )][2];
	Slope = "Slope = " || format(Slope,"Fixed Dec",9,5);
	Text(Right Justified, textList, Slope);
);
report(biv)["Linear Fit"] << visibility("collapse");
report(biv)[PictureBox(2)] << visibility("collapse");

Documentation for the Graphic additions, and the Display Tree manipulationq can be found in the Scripting Guide and the Scripting Index

Re: Need help writing optimal code for extracting R^2 and Slope from Bivariate Fit

HadleyMyers — Thu, 02 Jun 2022 07:56:28 GMT

Names Default To Here(1);

dt=Open("$SAMPLE_DATA/Tablet Production.jmp");

biv=dt<<Bivariate(
	Y( :Dissolution ),
	X( :Mill Time ),
	Fit Line( {Line Color( {212, 73, 88} )} )

);

slope=report(biv)[NumberColBox(13)][2];
r2=report(biv)[NumberColBox(1)][1];

slope=round(slope,3);
r2=round(r2,3);


Eval(Parse(Eval Insert("biv<<SendToReport(
		Dispatch(
			{},
			\!"Bivar Plot\!",
			FrameBox,
			Add Text Annotation(
				Text( \!"Rsq=^r2^
Slope=^slope^\!" ),
				Fixed Size( 0 ),
				Text Box( {2, 16, 100, 70} ),
				Filled( 0 )
			)
		)
	);")));