Problem with Nonlinear Least Squares fitting

5 次查看（过去 30 天）
Jason Yee
Jason Yee 2022-6-13
评论： Jason Yee 2022-7-4
I am trying to create a script that will fit some scientific data to the function
where a,b, and c are the fitting parameters. My problem is that the script does not seem to seek better parameters, but I can do the fitting in a different software. What is wrong with my code? My code is below, and I am very new to MatLab. Thank you.
x = A(389:401, 6);
y = A(389:401, 5);
axis([0 1010 0 1.6E-7])
% hold on
plot(x,y,'ro')
title('Data points')
% hold off
ft = fittype('(a/2)*(sin(b*pi/2)/(cosh(b*log(2*pi*x*c))+cos(b*pi/2)));');
fo = fitoptions('Method','NonlinearLeastSquares',...
'Lower',[-1,-1,-1],...
'Upper',[1,1,1],...
'StartPoint',[5e-7 0.63 9e-5]);
[f2,gof,output]=fit(x,y,ft,fo);
f2
gof.rsquare
figure
plot(f2,x,y)

请先登录，再回答此问题。

回答（1 个）

Sam Chak
Sam Chak 2022-6-14
编辑：Torsten 2022-6-14
@Jason Yee
I don't see anything technically wrong with code. Probably the choices of the lower and upper bounds are 'too vast'. Your data lies in this region?
axis([0 1010 0 1.6E-7])
Does your data look like these two curves?
x = linspace(-pi, pi, 3601);
y1 = (1/2)*(sin(1*pi/2)./(cosh(1*log(2*pi*x*1))+cos(1*pi/2)));
y2 = (-1/2)*(sin(-1*pi/2)./(cosh(-1*log(2*pi*x*-1))+cos(-1*pi/2)));
plot(x, [y1' y2'])
  1 个评论
Jason Yee
Jason Yee 2022-7-4
Hello
I tried narrowing the constraints on my starting values.
ft = fittype('(a/2)*(sin(b*pi/2)/(cosh(b*log(2*pi*x*c))+cos(b*pi/2)));');
fo = fitoptions('Method','NonlinearLeastSquares',...
'Lower',[0,0,0],...
'Upper',[1e-6,1,1e-4],...
'StartPoint',[5.4e-7 0.67 9e-5]);
[f2,gof,output]=fit(x,y,ft,fo);
f2
gof.rsquare
figure
plot(f2,x,y)
Do you recommend the lsqcurvefit function? Thank you.

请先登录，再进行评论。

类别

在 Help Center 和 File Exchange 中查找有关 Linear and Nonlinear Regression 的更多信息

标签

产品


版本

R2022a

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by