turn on suggestions

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

Showing results for

- JMP User Community
- :
- Discussions
- :
- Discussions
- :
- How is the Smooth Line in the Graph Builder determined

Topic Options

- 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

May 26, 2010 8:24 AM
(6571 views)

Can anyone tell me how the smooth line is generated in the Graph Builder? I assume is some sort of running average, but anyone know how many values or what range its pulling to make the average?

1 ACCEPTED SOLUTION

Accepted Solutions

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

May 26, 2010 1:10 PM
(10858 views)

I raised this question with JMP support and received the following which I have slightly paraphrased:

The smoother utilized in the Graph Builder platform is a cubic spline with a lambda of 0.05 and standardized X values. The same spline can be obtained through the Bivariate platform by selecting Analyze > Fit Y by X, supplying the Y and X variables, and clicking OK. On the resulting Bivariate report window, select Fit Spline > Other from the popup menu. Then supply a smoothness parameter (lambda) of 0.05, and check the Standardize X box.

This information is from the JMP Statistics and Graphics Guide for version 8.0.2 and this info is found on pages 929-930 in Chapter 43. This is in the section on "Changing the Graph Element".

The "standardization of X" has the same effect as subtracting the mean and dividing by the standard deviation for the X variable and then fitting the spline. The only difference is that in Graph Builder and the Fit Y by X platform, the data is still plotted on the original scale rather than the standardized scale. Using standardized X values seems to smooth out the fit a little

The smoother utilized in the Graph Builder platform is a cubic spline with a lambda of 0.05 and standardized X values. The same spline can be obtained through the Bivariate platform by selecting Analyze > Fit Y by X, supplying the Y and X variables, and clicking OK. On the resulting Bivariate report window, select Fit Spline > Other from the popup menu. Then supply a smoothness parameter (lambda) of 0.05, and check the Standardize X box.

This information is from the JMP Statistics and Graphics Guide for version 8.0.2 and this info is found on pages 929-930 in Chapter 43. This is in the section on "Changing the Graph Element".

The "standardization of X" has the same effect as subtracting the mean and dividing by the standard deviation for the X variable and then fitting the spline. The only difference is that in Graph Builder and the Fit Y by X platform, the data is still plotted on the original scale rather than the standardized scale. Using standardized X values seems to smooth out the fit a little

8 REPLIES

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

May 26, 2010 1:10 PM
(10859 views)

The smoother utilized in the Graph Builder platform is a cubic spline with a lambda of 0.05 and standardized X values. The same spline can be obtained through the Bivariate platform by selecting Analyze > Fit Y by X, supplying the Y and X variables, and clicking OK. On the resulting Bivariate report window, select Fit Spline > Other from the popup menu. Then supply a smoothness parameter (lambda) of 0.05, and check the Standardize X box.

This information is from the JMP Statistics and Graphics Guide for version 8.0.2 and this info is found on pages 929-930 in Chapter 43. This is in the section on "Changing the Graph Element".

The "standardization of X" has the same effect as subtracting the mean and dividing by the standard deviation for the X variable and then fitting the spline. The only difference is that in Graph Builder and the Fit Y by X platform, the data is still plotted on the original scale rather than the standardized scale. Using standardized X values seems to smooth out the fit a little

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

May 26, 2010 1:17 PM
(6334 views)

thanks, thats exactly the information I was looking for :)

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

Oct 30, 2017 2:36 AM
(5536 views)

Hi,

Is there a way to extract the equations of the smoothed curves ?

Thanks

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

Oct 30, 2017 10:46 AM
(5516 views)

@samir here are two posts on cubic splines.

this one is probably what you want (thanks @Duane_Hayes):

this one shows another way to get the coefficients; I used the smoothed spline values to control the color.

Craige

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

Oct 31, 2017 6:27 AM
(5487 views)

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

Feb 22, 2018 1:43 PM
(3730 views)

I see the option of cubic splines is available in JMP at graph builder with categorical variables. For example, when you test the moderating(interacting) effect of a categorical variable on the effect of a categorical variable to a categorical outcome. Does this make sense? Is that possible?

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

Feb 23, 2018 7:29 AM
(3707 views)

I can make sense in cases like this where the response has two levels and the factor is ordinal. In that case, the Y reflects the proportion of the two values and you get of the trend for how that proportion changes with the GPA. Your "Missing" value on the X axis, however, is categorically different and should be plotted separately.

A more common view for categorical data would be a stacked bar of the counts or proportions for each outcome for each GPA value.

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

Feb 23, 2018 7:43 AM
(3703 views)

Thank you very much for your prompt reply and heartening answer.