topic Graph Builder - small and big subset in Discussions

Graph Builder - small and big subset

hogi — Fri, 09 Jun 2023 16:02:50 GMT

I have some (outlier) data points and want to compare the individual values with 2 reference sets.

ideal: points for the small set and violin for the references.

Can I specify this in Graph Builder?

workaround 1: fill the x axis twice

workaround 2: ... with a second question
use points (for everything) - and packed jitter to get a feeling for the distribution.
But graph Builder is a little lazy when calculating packed jitter positions for larger data sets.

Can i change this behavior?

generatePlot=Function ({nRows},{Default Local},
	dt = New Table();
	dt << add rows( nRows );
	dt:Column1 << set name( "group" ) << Set Modeling Type( "Nominal" ) <<
	Formula( tmp=Random uniform( 2 );If(tmp<0.001,0,tmp < 0.5, 1,2) );
	
	dt << New Column( "select", Formula( Random Uniform( ) ) );
	dt << New Column( "rnd", Formula( Random Normal( ) ) );

Graph Builder(
	Size( 523, 411 ),
	Show Control Panel( 0 ),
	Graph Spacing( 4 ),
	Variables(
		X( :group ),
		Y( :rnd ),
		Overlay( :group )
	),
	Elements( Position( 1, 1 ), Points( X, Y, Legend( 17 ), Jitter( "Packed" ) ) )
));


generatePlot(100000); // how it should look like
generatePlot(500000); // larger data set
generatePlot(1000000); // even larger data set

Re: Graph Builder - small and big subset

ih — Tue, 06 Dec 2022 22:14:18 GMT

You could build that in one chart by creating separate x or y columns for each type of graph you want, and then limit which variables are used in each feature.

Code to reproduce:

Names default to here(1);

generatePlot=Function ({nRows},{Default Local},
	dt = New Table();
	dt << add rows( nRows );
	dt:Column1 << set name( "group" ) << Set Modeling Type( "Nominal" ) <<
	Formula( tmp=Random uniform( 2 );If(tmp<0.001,0,tmp < 0.5, 1,2) );
	
	dt << New Column( "select", Formula( Random Uniform( ) ) );
	dt << New Column( "rnd", Formula( Random Normal( ) ) );
	dt << New Column("Reference", Numeric, "Continuous", Format("Best", 12), Formula(If(:group == 0, Empty(), :rnd)));
	dt << New Column("Value", Numeric, "Continuous", Format("Best", 12), Formula(If(:group == 0, :rnd)));

	dt << Graph Builder(
		Size( 840, 589 ),
		Show Control Panel( 0 ),
		Variables( X( :group ), Y( :Reference ), Y( :Value, Position( 1 ) ) ),
		Elements(
			Points( X, Y( 2 ), Legend( 19 ) ),
			Box Plot( X, Y( 2 ), Legend( 22 ) ),
			Contour( X, Y( 1 ), Legend( 23 ) )
		),
		SendToReport(
			Dispatch( {}, "graph title", TextEditBox, {Set Text( "rnd vs. group" )} ),
			Dispatch( {}, "Y title", TextEditBox, {Set Text( "rnd" )} )
		)
	)	
);


//All three should look the same
generatePlot(100000);
generatePlot(500000);
generatePlot(1000000);

Re: Graph Builder - small and big subset

hogi — Tue, 06 Dec 2022 22:32:49 GMT

nice :)