topic Means Comparison Oneway Analysis ... how to dynamically order side-by-side boxplots by mean value left to right in Discussions

Means Comparison Oneway Analysis ... how to dynamically order side-by-side boxplots by mean value left to right

galactus3000 — Fri, 04 Oct 2019 18:59:19 GMT

Looking for an automated way to automatical order side-by-side box plots by mean value left to right ... and by dynamically I mean being able to seamlessly do this on the fly in response to change in Local Data Filter and/or Column Switcher. Thanks in advance!

Re: Means Comparison Oneway Analysis ... how to dynamically order side-by-side boxplots by mean value left to right

gzmorgan0 — Tue, 08 Oct 2019 07:34:39 GMT

@galactus3000,

Which JMP platform are you using Fit Y by X or something else? Off the top of my head, if using Fit Y by X, I would use a custom script to add the data filter and add to it a Data Filter Handler. That is a bit involved.

If you are using Graph Builder to draw your boxplots, you can right click in the X-Axis (group axis) and select which Order by Statistic, default is the mean, and Select which Y variable to order by either ascending or desending.

The script below opens the JMP sample data table SATByYear.jmp and creates a Graph Builder boxplot of SAT Math vs. State with a Local Data Filter for Region. Now select one or several regions and ordering is updated automatically. Change from descending to ascending. You can do all of this voa point & click.

I hope this meets your needs.

Names Default to Here(1);
dt = Open("$Sample_data/SATByYear.jmp");

gb = dt << Graph Builder(
	Size( 534, 464 ),
	Show Control Panel( 0 ),
	Variables(
		X( :State, Order By( :SAT Math, Descending, Order Statistic( "Mean" ) ) ),
		Y( :SAT Math )
	),
	Elements( Box Plot( X, Y, Legend( 6 ) ) ),
	Local Data Filter(
		Add Filter(
			columns( :Region ),
			Display( :Region, Size( 160, 120 ), List Display )
		)
	)
);

Re: Means Comparison Oneway Analysis ... how to dynamically order side-by-side boxplots by mean value left to right

galactus3000 — Tue, 08 Oct 2019 12:56:46 GMT

yes using Fit Y by X just like I was taught in my first-ever JMP-based statistics course in 1997 in Portland Oregon