Subscribe Bookmark RSS Feed

X and Y error bars in X-Y Plot

robhanssen

Community Trekker

Joined:

Aug 27, 2014

For physical property correlation tests, we often measure two properties on a series of samples differentiated by a varying concentration of an additive. These two property measurements are repeated 5 times for each sample (at a set concentration). What I would like to do, is plot the two properties on a scatter plot and show the error bars for both properties by concentration.

 

In Excel, I would calculate mean, std.dev and the confidence interval. The confidence interval can then be used to set the size of the error bars in the X and Y direction.

 

JMP is bound to have a much better and more statistically sound mechanism for this, I hope? I've added some example files, to illustrate my intention. These are randomly generated data.

1 ACCEPTED SOLUTION

Accepted Solutions
ms

Super User

Joined:

Jun 23, 2011

Solution

With some effort it can be done with Graph Builder in JMP 13 by plotting Y positioned at mean_X and X positioned at mean_Y on top of eachother but with different response axis settings. In the example below locally tranformed columns are used within Graph Builder (i.e. no need to change the data table). 

 

X-Y.png 

 

Here the script to reproduce it:

Graph Builder(
    Variables(
        X(:Property1),
        X(Transform Column("mean_Prop1", Formula(Col Mean(:Property1, :concentration))), Position(1)),
        Y(:Property2),
        Y(Transform Column("mean_Prop2", Formula(Col Mean(:Property2, :concentration))), Position(1)),
        Overlay(:concentration)
    ),
    Elements(
        Points(X(1), Y(2), Response Axis("X"), Summary Statistic("Mean"), Error Bars("Confidence Interval")),
        Points(X(2), Y(1), Response Axis("Y"), Summary Statistic("Mean"), Error Bars("Confidence Interval"))
    )
);

 

 

 

 

 

3 REPLIES
Jeff_Perkinson

Community Manager

Joined:

Jun 23, 2011

How about a bivariate density elipse for each concentration?

 

JMPScreenSnapz074.png

You can get this using Fit Y by X with your two measurements.

 

Then choose Group By... from the red triangle hotspot and choose Concentration as your grouping column.

JMPScreenSnapz072.png

Then add a density elipse.

 

JMPScreenSnapz073.png

-Jeff
ms

Super User

Joined:

Jun 23, 2011

Solution

With some effort it can be done with Graph Builder in JMP 13 by plotting Y positioned at mean_X and X positioned at mean_Y on top of eachother but with different response axis settings. In the example below locally tranformed columns are used within Graph Builder (i.e. no need to change the data table). 

 

X-Y.png 

 

Here the script to reproduce it:

Graph Builder(
    Variables(
        X(:Property1),
        X(Transform Column("mean_Prop1", Formula(Col Mean(:Property1, :concentration))), Position(1)),
        Y(:Property2),
        Y(Transform Column("mean_Prop2", Formula(Col Mean(:Property2, :concentration))), Position(1)),
        Overlay(:concentration)
    ),
    Elements(
        Points(X(1), Y(2), Response Axis("X"), Summary Statistic("Mean"), Error Bars("Confidence Interval")),
        Points(X(2), Y(1), Response Axis("Y"), Summary Statistic("Mean"), Error Bars("Confidence Interval"))
    )
);

 

 

 

 

 

robhanssen

Community Trekker

Joined:

Aug 27, 2014

Thanks for taking the time to help me!

 

Is there an obvious way to generate this script through the Graph Builder application? I might invest time to generate scripts this way, but I'd like to convey this to my co-workers as well.