Flagging Site to site variations

[#1]

Hi,

Is there a way to flag test parameters with site-to-site variations? 

I am aware of looking at the variability chart, but I am exploring ways to identify statistically and implement using jsl which would make analysis faster.

The thought is to create a column that indicates if the test parameter has site-to-site variations. Any suggestions?

dt = Open( "$SAMPLE_DATA/Semiconductor Capability.jmp" );

Thanks,  

[#2]

Have you tried Response Screening (https://www.youtube.com/watch?v=IoMYIGPuDKk), https://www.jmp.com/support/help/en/17.2/index.shtml#page/jmp/response-screening.shtml#?

[#3 · Accepted solution]

Not sure if I understand exactly what you want, but you could use the builtin Oneway platform and extract the means comparison results? Something like this:

names default to here(1);
dt = Open( "$SAMPLE_DATA/Semiconductor Capability.jmp" );
dt << New Column( "PM_L1_Modified", set formula( If( :SITE == 1, :PM_L1 + 1, :PM_L1 ) ) );
cols1 = dt << get column names( Continuous );

For( j = 1, j <= n items( cols1 ), j++, 
	ow = dt << Oneway(
		Y( cols1[j] ),
		X( :SITE ),
		All Pairs( 1 ),
		Comparison Circles( 0 ),
		Invisible
	);

	level1 = (Report( ow )["Means Comparisons", "Comparisons for all pairs using Tukey-Kramer HSD", "Ordered Differences Report",
	String Col Box( "Level" )]) << get();
	level2 = (Report( ow )["Means Comparisons", "Comparisons for all pairs using Tukey-Kramer HSD", "Ordered Differences Report",
	String Col Box( "- Level" )]) << get();
	pval = (Report( ow )["Means Comparisons", "Comparisons for all pairs using Tukey-Kramer HSD", "Ordered Differences Report",
	String Col Box( "p-Value" )]) << get();

	ow << close window();
	p_count = 0;
	For( i = 1, i <= N Items( pval ), i++,
		If( pval[i] < 0.05,
			text = Substitute( "For __col__, sites __l1__ and __l2__ are statistically different",
				Expr( __col__ ), cols1[j],
				Expr( __l1__ ), level1[i],
				Expr( __l2__ ), level2[i]
			);
			Print( text );
			p_count++;
			// Do stuff here
		)
	);
	if( p_count == 0, print( char( cols1[j] ) || ": no statistically different sites" ) );
);

Then you'd get output like this:

...
"PM_L1: no statistically different sites"
"P1: no statistically different sites"
"M1: no statistically different sites"
"For PM_L1_Modified, sites 1 and 5 are statistically different"
"For PM_L1_Modified, sites 1 and 2 are statistically different"
"For PM_L1_Modified, sites 1 and 4 are statistically different"
"For PM_L1_Modified, sites 1 and 3 are statistically different"

I had to create a new column as none of the parameters in Semiconductor Capability seem to have a SITE dependancy! 

You can then do something interesting with the parameter and sites which are found to have statistically significant differences.

This is a very simple solution and could be optimised, or there may be a better JMP built-in way to get this info.

[#4]

This is what I am looking for. Thanks @matth1