fitnlm problem- the SE is very small
1 view (last 30 days)
Show older comments
Hello all, I am using 'fitnlm' to fit a non-linear model, however, the SE in mdl.Coeffficients is very small, in the order of e-16~e-30.
![E{7Q`2A20)@LD1R`_4B18_U.png](https://www.mathworks.com/matlabcentral/answers/uploaded_files/201506/E{7Q`2A20)@LD1R`_4B18_U.png)
The model for sure cannot fit perfectly to the data, if I feed the data into the curve fitting toolbox using the same initial value, I can get reasonable boundary for the model coefficients.
General model:
f(x) = 1-a*x^b-c*x^d
Coefficients (with 95% confidence bounds):
a = 0.002713 (0.001471, 0.003956)
b = 0.6173 (0.5348, 0.6998)
c = 9.554e-20 (-6.96e-19, 8.871e-19)
d = 6.375 (5.121, 7.629)
Their goodness of fittting are both good.
Did anyone meet similar problems? Thank you.
0 Comments
Answers (0)
See Also
Categories
Find more on Interpolation in Help Center and File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!