Huge Confidence Interval With predint

4 次查看（过去 30 天）
Samuel Salander
Samuel Salander 2021-3-24
回答： Aditya Patil 2021-3-31
I'm running into a confidence interval on a fit of some data. This is the code I'm using:
x0 = [1 1 1];
fitfun = fittype( @(a,b,c,x) a*x.^b+c)
[fitted_curve,gof] = fit(x,y,fitfun,'StartPoint',x0,'Weight', w, 'Lower', [0,-3,0],'Upper',[1000,0,1])
coeffvals = coeffvalues(fitted_curve);
p = predint(fitted_curve,x,0.001,'Observation','off');
Where w is calculated from inverse cube weighting:
The fit works well for this very small confidence interval (.1 % for .001). When I try to make a more meaningful fit and use 95% confidence, I get this:
Is there a way I can get a closer result for this 95% interval? predint must use "Observation" and "off" (changing to "functional" makes interval smaller, but it's not what I'm looking for).
Thanks so much for your help.

请先登录，再回答此问题。

回答（1 个）

Aditya Patil
Aditya Patil 2021-3-31
As per my understanding, you want to get a fit with lower confidence interval.
For this, you need to do one or more of the following,
  1. Modify data. Add new data, remove outliers, and converting between data representations
  2. Change fit function. Use a function with more degrees.
  3. Remove upper and lower limits on the coefficients.

类别

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

标签

Community Treasure Hunt

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

Start Hunting!

Translated by