topic Re: script for creating a table of p values in Discussions

script for creating a table of p values

TRR21 — Fri, 04 May 2018 16:58:54 GMT

I have a large table of 40 rows (4 groups of 10 samples) and 10000 columns of values—

1. Is there a way for me to quickly run t-tests/ANOVA comparing particular group against another group, for each of those columns and list the p-values in a table to quickly sort through to understand the ones that are significantly different?

2. Also, is it possible to set the p-value is such test to correct for multiple correction?

Thanks

Re: script for creating a table of p values

txnelson — Fri, 04 May 2018 17:55:03 GMT

Run the Response Screening Platform and it will do that for you

Analyze==>Screening==>Response Screening

Re: script for creating a table of p values

TRR21 — Fri, 04 May 2018 21:38:53 GMT

Thanksf or the fast response - wouldn't have guessed this is what it would do.

Is there a way to trigger this from a script, selecting only particular rows for this analysis, based on data in couple other columns (ie. some way to subset/filter), without having to make that table first?

Re: script for creating a table of p values

txnelson — Sat, 05 May 2018 01:51:08 GMT

You can use the ability to set the Row State for given rows to be Excluded from the subsequent analysis.

Is this what you are trying to do?

Re: script for creating a table of p values

TRR21 — Mon, 07 May 2018 18:10:33 GMT

Thanks, excluding all the rows I DON'T want does get me where I need to go.

Re: script for creating a table of p values

TRR21 — Mon, 07 May 2018 21:00:02 GMT

Is there a way to choose between parametric vs. non-parametric/rank test to be used in this? None of the options available in the dialog seem to apply. Thanks.

Re: script for creating a table of p values

TRR21 — Thu, 28 Jun 2018 00:56:14 GMT

Is there a way to force the system to used a paired test to generate the p-vals? Thanks.

Re: script for creating a table of p values

txnelson — Thu, 28 Jun 2018 01:45:48 GMT

JMP 13 and JMP 14

Analyze==>Specialized Modeling==>Matched Pairs

JMP 12 and earlier

Analyze==>Matched Pairs

For future reference, this information can be found in the Statistics Index, searching on "paired"

Help==>Statistics Index

Re: script for creating a table of p values

TRR21 — Sat, 07 Jul 2018 00:12:08 GMT

Thanks - but am not sure if this is the one I want. What I want is more like, what I get if I use:
Analyze==>Fit Y by X; and from red triangle pull down, use "Matching column"
This gives me the pre-vs-post paired comparison I am looking for, but would like to run this for the 20K columns as mentioned above.

Re: script for creating a table of p values

txnelson — Sat, 07 Jul 2018 02:25:22 GMT

Here is a sample script that might be close to what you want. I had to guess a bit at what matching you were going to do. But, it creates a table of output for all 127 columns in the data table.

Names Default To Here( 1 );

// Create a data table somewhat similar to your description
dtorig = Open( "$SAMPLE_DATA\semiconductor capability.jmp" );

dt = dtorig << Subset( Sample Size( 10 ), Selected columns only( 0 ), Stratify( :SITE ) );

Close( dtorig, nosave );

// Get all of the tests
colNamesList = dt << get column names( continuous );

// In my example, I am using the first column as the matching column
// so remove it from the list
colNamesList = Remove( colNamesList, 1, 1 );

// Run the Fit Y by X(Oneway)
fg = Fit Group(
	Oneway(
		Y( Eval( colNamesList ) ),
		X( :SITE ),
		Matching Column( :NPN1 ),
		X Axis Proportional( 0 ),
		Matching Lines( 1 ),
		SendToReport(
			Dispatch( {}, "Oneway Plot", FrameBox, {Grid Line Order( 2 ), Reference Line Order( 3 )} ),
			Dispatch( {}, "Matching Fit", OutlineBox, {Close( 0 )} )
		)
	)
);

// Create the output table
Report( fg )[Table Box( 3 )] << make combined data table;

// Close the report output
Report( fg ) << close window;

Re: script for creating a table of p values

taxkourel — Fri, 19 Aug 2022 22:27:58 GMT

How do we know if the significantly different values are increased/decreases and how do we know the baseline group?

Re: script for creating a table of p values

txnelson — Fri, 19 Aug 2022 23:06:29 GMT

The Oneway Platform allows for the display of the group means. The sample script does not capture that data, but it could have that information included.

Concerning the identification of the Baseline Group. the assumption is that either the data would contain appropriate labeling, or that the analyzer of the data would know that information.