How could I put FIT parameters' errors in a variable?

11 次查看(过去 30 天)
Hi, i know that I could do:
fuz=fittype('(a*x)/(1+b*(x^2))');
d=fit((final(:,8)),(final(:,6)),fuz,'StartPoint',[r1,s1]);
coeffi=coeffnames(fuz);
coefvv=coeffvalues(d);
g=coefvv'
and in g(1) there's the value of a and in g(2) the value of b.
But if I do:
fuz=fittype('(a*x)/(1+b*(x^2))');
[val1,val2,val3]=fit((final(:,8)),(final(:,6)),fuz,'StartPoint',[r1,s1]);
disp(val1);
I have also the error bars on the parameters...How could I put them into variables?

采纳的回答

Gabriele
Gabriele 2013-5-29
if I do disp(val1) I display in the command window:
General model:
val1(x) = (a*x)/(1+b*(x^2))
Coefficients (with 95% confidence bounds):
a = 0.0002983 (-0.0002809, 0.0008774)
b = 1.93e-009 (-3.784e-008, 4.17e-008)
how can I put the two numbers in (-3,784e-008, 4.17e-008) in a variable?

更多回答(1 个)

Gabriele
Gabriele 2013-5-31
the solution is reported here http://www.mathworks.it/it/help/curvefit/confint.html I hope it would be useful for anyone in the future...

类别

Help CenterFile Exchange 中查找有关 Measurements and Feature Extraction 的更多信息

Community Treasure Hunt

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

Start Hunting!

Translated by