Determining function for smoothing spline

Question

Gavin Seddon 2017-3-27

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/332123-determining-function-for-smoothing-spline

评论： Image Analyst 2022-4-4

Hello, my data when plotted resembles a plot of a sine function and consequently a polynomial curve will not converge. Therefore I am using a smoothing spline. Is it possible to determine the cubic function associated with this?

GS.

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Stephen23 2017-3-27

编辑：Stephen23 2017-3-27

The function spline returns a structure which encodes the spline.

And read this:

https://www.mathworks.com/matlabcentral/answers/325706-natural-spline-unmkpp-gives-wrong-answer

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Image Analyst 2017-3-27

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/332123-determining-function-for-smoothing-spline#answer_260528

在 MATLAB Online 中打开

The equations of the cubics between each pair of knots is given in the coefs field. See how I took the demo script from the help and printed them out:

x = -4:4;
y = [0 .15 1.12 2.36 2.36 1.46 .49 .06 0];
cs = spline(x,[0 y 0]);
xx = linspace(-4,4,101);
plot(x,y,'o',xx,ppval(cs,xx),'-');
% Now print out the coefficients
coefficients = cs.coefs
fprintf('The equation for the different segments are:\n');
for k = 1 : size(coefficients, 1)
  fprintf('y = %7.4f * x^3 + %7.4f * x^2 + %7.4f * x + %7.4f\n', coefficients(k,:));
end

In the command window, you will see:

coefficients =
          0.20344440353461       -0.0534444035346097                         0                         0
       -0.0903332106038293          0.55688880706922          0.50344440353461                      0.15
        -0.392111561119293         0.285889175257731          1.34622238586156                      1.12
         0.148779455081001        -0.890445508100147         0.741666053019146                      2.36
         0.136993740795287        -0.444107142857143        -0.592886597938144                      2.36
          0.13324558173785       -0.0331259204712812         -1.07011966126657                      1.46
       -0.0599760677466861         0.366610824742268        -0.736634756995581                      0.49
       -0.0633413107511046         0.186682621502209        -0.183341310751105                      0.06
The equation for the different segments are:
y =  0.2034 * x^3 + -0.0534 * x^2 +  0.0000 * x +  0.0000
y = -0.0903 * x^3 +  0.5569 * x^2 +  0.5034 * x +  0.1500
y = -0.3921 * x^3 +  0.2859 * x^2 +  1.3462 * x +  1.1200
y =  0.1488 * x^3 + -0.8904 * x^2 +  0.7417 * x +  2.3600
y =  0.1370 * x^3 + -0.4441 * x^2 + -0.5929 * x +  2.3600
y =  0.1332 * x^3 + -0.0331 * x^2 + -1.0701 * x +  1.4600
y = -0.0600 * x^3 +  0.3666 * x^2 + -0.7366 * x +  0.4900
y = -0.0633 * x^3 +  0.1867 * x^2 + -0.1833 * x +  0.0600

4 个评论
显示 2更早的评论隐藏 2更早的评论

Nilanshu Mahant 2022-4-4

Hi, did you manage to find any equation. because I'm also working on same situiation. I'm using curve fitting tool to identify the corelation between different parameters and usning spline. My condition is match with the spline. But, now I'm wondering how to create a equation from that.

Could you please suggest?

Image Analyst 2022-4-4

@Nilanshu Mahant not sure what you mean. The cubic spline determines a cubic polynomial for each segment between two points. For example in my answer above I gave all the equations between all the data points. There is not really any need to create dozens or hundreds of equations. Just use the spline function. By its very nature there is no one, overall cubic function for all of your dozens or hundreds of points. You get an equation between each pair of points. If you need some overall analytical formula instead of doing it numerically with spline, then I suggest you use the theoretical formula for your physical process and then use fitnlm() to get the parameters for your formula. If you need more help, start your own question and attach your data and code to read it in and plot it, and what your desired output would be.

请先登录，再进行评论。