cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
JMP is taking Discovery online, April 16 and 18. Register today and join us for interactive sessions featuring popular presentation topics, networking, and discussions with the experts.
Choose Language Hide Translation Bar
marcokunzelmann
Level III

cubic spline regression

Hello everybody,

I'm looking for a solution to regress a bivariate data set with a cubic spline regression which has exactly 15 knots. It would be great if someone has a solution for me!

Greetings

1 ACCEPTED SOLUTION

Accepted Solutions
marcokunzelmann
Level III

Re: cubic spline regression

Hello and thanks for your answer! Exactly something like that, but the question is, does lambda represent the amount of knots for my cubic spline?

Edit// Ok I have read now, that your suggestion is a smoothing spline. Smoothing splines are not calculating with knots instead you have penalty function with the factor lambda. So the question is, is there another function in JMP that I can calculate the splines with knots?

By the way is there a nice feature how I can extract the formula of the spline regression and calculate "new" height values by just input the weight values?

Edit// I think I found a solution. The knotted spline effect should be fine for me, it is hidden under Analyze->FitModel->Attributes

To extract formulas and calculate new values, just click on the red red triangle->save columns->prediction formula

In the next step just write down the new x-values in the same column were the old x-values are already. The predicted values are displayed in the new column!

View solution in original post

3 REPLIES 3
louv
Staff (Retired)

Re: cubic spline regression

Something like this

Bivariate(

  Y( :height ),

  X( :weight ),

  Fit Spline( 15, {Line Color( {208, 64, 86} )} )

);


8930_Screen Shot 2015-06-15 at 9.04.25 PM.png

marcokunzelmann
Level III

Re: cubic spline regression

Hello and thanks for your answer! Exactly something like that, but the question is, does lambda represent the amount of knots for my cubic spline?

Edit// Ok I have read now, that your suggestion is a smoothing spline. Smoothing splines are not calculating with knots instead you have penalty function with the factor lambda. So the question is, is there another function in JMP that I can calculate the splines with knots?

By the way is there a nice feature how I can extract the formula of the spline regression and calculate "new" height values by just input the weight values?

Edit// I think I found a solution. The knotted spline effect should be fine for me, it is hidden under Analyze->FitModel->Attributes

To extract formulas and calculate new values, just click on the red red triangle->save columns->prediction formula

In the next step just write down the new x-values in the same column were the old x-values are already. The predicted values are displayed in the new column!

triunk
Level I

Re: cubic spline regression

I see the option of cubic splines is available in JMP at graph builder when you test the moderating effect of a categorical variable on the effect of a categorical variable to the outcome which happens to be... you guessed it...also categorical variable. Does this make sense? Is that possible?