Subscribe Bookmark RSS Feed

Polynomial Fit Parameter Term Name Changes: Bug?

robot

Community Trekker

Joined:

Feb 27, 2012

Hi,

I am writing a script that generates a second degree polynomial fit and then uses the parameter estimates in further calculations.  My problem is the second order parameter estimate term name changes as a function of the input values, when it seems to me it should not.  This term name change causes subsequent calculations to fail.

For example, the script below will generate a second degree polynomial fit with the following terms:

Intercept

height

(height-62.55)^2

My question is, why is the second order term name "(height-62.55)^2"?  Shouldn't the term name be "height^2"?  This name will change as a function of the input values.  Is this a bug?

Any insights or workarounds are appreciated.

Thanks!

/*

bigClass=open("$SAMPLE_DATA/Big Class.JMP");

bigClass << Bivariate(

  Y( :weight ),

  X( :height ),

  Fit Polynomial( 2, {Line Color( {213, 72, 87} )} )

);

*/

1 ACCEPTED SOLUTION

Accepted Solutions
Solution

By default JMP Bivariate Fit Polynomial generates a model with the terms of order 2 and greater centered around the mean of the X variable. It names the coefficients of those terms in a way that communicates what it did.

Instead use Fit Special, make sure neither X nor Y is transformed, change the degree to "2 quadratic" and uncheck "Centered Polynomial".

3 REPLIES
robot

Community Trekker

Joined:

Feb 27, 2012

Note: I am using JMP10.

Solution

By default JMP Bivariate Fit Polynomial generates a model with the terms of order 2 and greater centered around the mean of the X variable. It names the coefficients of those terms in a way that communicates what it did.

Instead use Fit Special, make sure neither X nor Y is transformed, change the degree to "2 quadratic" and uncheck "Centered Polynomial".

robot

Community Trekker

Joined:

Feb 27, 2012

Works great.  Thanks!