nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

How to Plot graph with reference lines for each element of table summary

spinup

Hi All,

I am trying to plot graphs (script , JMP 13) with reference lines for each element of my table summary (graph of score vs # try for each Name+Grade, then add reference line of 25 and 75 percentile of Score on x axis). Would appreciate any comment

Name	Grade	Score	# try
name1	A	80	2
name1	A	82	3
name1	B	98	3
name1	C	86	2
name2	A	88	3
name3	A	90	2
name3	A	88	1
name3	A	94	3
name4	B	96	2
name4	B	82	4
name4	A	100	1
name4	C	92	2
name5	A	84	3

Accepted answers

spinup

Thanks @Craige_Hales for the link, but my main difficulty is how to program to output the individual graphs as each have their own reference lines, more clearly, I am wondering how to generate graphs for each color in this data table :

jmp question-2.jpg

Thanks!

SDF1

Hi @spinup,

I think I understand what you want. But, I'm also a little confused: are you wanting to summarize or just graph the 9 separate bivariate graphs?

Working off your original data table, you can easily create all 9 graphs if you do a bivariate fit, e.g.:

Bivariate( 
  Y( :Score ), X( :Name( "# try" ) ), 
  By( :Name, :Grade )
);

This will generate your nine graphs that you had colored in your previous post.

Then, what you'll probably have to do is use a For Loop(s) to generate the quantile lines (or shaded area).

The script for the first graph would be something like:

Bivariate(
	Y( :Score ),
	X( :Name( "# try" ) ),
	Where( :Name == "name1" & :Grade == "A" ), // this would be selecting 
	SendToReport(
		Dispatch(
			{},
			"2",
			ScaleBox,
			{Min( 75 ), Max( 100 ), Inc( 5 ), Minor Ticks( 1 ), // this sets the appearance of the y-axis
			Add Ref Line( 83, "Solid", "Blue", "25th quantile", 1 ),
			Add Ref Line( 95, "Solid", "Blue", "75th quantile", 1 )}
		)
	)
);

In fact, if you set up a list of the names NameList={"Name1", "Name2"...}, or pull it from the column elements, then you can have a very short script by putting in For Loops and just us reference the NameList[i] element. This would be also the case for the Where(:Name==NameList[i] & Grade == GradeList[i]).

Another option would be to inspect the tree structure of the output from the first script I included above and then you'd reference the AxisBox(i) for each of the y-axes for the 9 graphs. The y-axes are all odd, so if you do a For Loop, then you'd have to do i++ so that it iterates from 1 to 3 to 5 and so on in the loop.

There's some really good stuff in the link that @Craige_Hales provided.

For your specific issue, I think you'll definitely need to do some looping. Either that, or you can do the whole thing manually once and then just save the script to the data table (probably as a group), and then you can copy/paste that and save as a separate JSL file in a new script window. If you can figure out how to do it manually, JMP will write the JSL for you, and all you need to do is put it all together into a master JSL file that takes care of it all for you. The only issue there is if you decide to change names or column references or things like that. If you do, then you'll need to start using local variables.

Hope this helps!,

All comments

SDF1

Hi @spinup,

I'm not sure I understand entirely what you want to do, but you might have a hard time scripting the whole thing entirely. If you are keeping the data table around for a while, or load it in, you might want to do the analyses separately and then save them to the data table.

If you do want to script the whole thing, you'll need to still run the analyses to determine the 25/75 quantiles, etc.

When you say you're graphing "score vs # try", do you mean when you do a "fit y by x" that score is "y" and "# try" is "x"?

Also, when you say table summary, do you mean you're finding the mean of score and #try while grouping by name and grade? Or, do you mean you're tabulating by name the grade and finding the mean of score & #try values?

You could do it a couple ways though. I took your table and brought it into JMP, and here's what I did (assuming I am thinking about it the same way as you):

Tabulate by Name (A, B, C are the # tries), this helps to get the means, but you can script this separately if needed.

With this at hand, I graphed it a couple different ways: bivariate fit or histogram (histogram already has the quartile info readily available from the box plot). and then you can add reference lines via the normal way.

For scripting all this out, be sure to use the red hot button to get to the actual JSL script to generate what you want. But again, if you save the script to the data table, then each time you open up the data table, you can click the green hot button and re-run it quickly.

Hope this helps!,

spinup

Thanks @SDF1 for your reply and sorry for not being very clear. Here I would like to explain a little more:
first of all, I am trying to write the script in a separate script window so I can run it independently, here is the first part of the script where I am trying to get the summary:

Names Default To Here( 1 );
dt = Current Data Table();
Summary = dt << Summary(
	Group( :Name, :Grade ),
	Median( :Score ), 
// other statistics can be added here//
	Link to Original Data Table( 0 )
);

the Output will look like this :

jmp question.jpg

Now , What I really want is the 9 graphs ( Score vs # try) (one for each row (having N Rows datapoints) ), and then use the statistics of each group to add to the reference lines ( all by scripting).

In your graph example, looks it is for name1 with any Grade.

Thanks

Craige_Hales

spinup

jmp question-2.jpg

Thanks!