Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

- JMP User Community
- :
- Discussions
- :
- X and Y error bars in X-Y Plot

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Feb 20, 2017 2:00 PM
(2411 views)

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

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Feb 20, 2017 4:36 PM
(4706 views)

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).

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

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Feb 20, 2017 2:47 PM
(2407 views)

How about a bivariate density elipse for each concentration?

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.

Then add a density elipse.

-Jeff

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Feb 20, 2017 4:36 PM
(4707 views)

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).

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"))
)
);
```

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Feb 21, 2017 11:47 AM
(2365 views)

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.