# How to obtain confidence level and coefficient of determination?

6 views (last 30 days)
Fabian Moreno on 7 Jul 2021
Commented: Fabian Moreno on 8 Jul 2021
Hello, I am comparing two regression models (linear fit and theil-sen). In the linear fit I can get confidence level (CI=95%) of the slop and intersection, and coefficient of determination (R2). But I don't know how to get the CI and R2 for theil sen model. I appreciate any help. Here is my code.
%Option MRL
mdl = fitlm(x, y,'linear','Intercept',true,'RobustOpts','on'); % Fit Data
B = mdl.Coefficients.Estimate;
B=[B(2) B(1)];% Coefficients
y_robus = polyval(B,x(:,1));
% Confidence intervals
CI = coefCI(mdl,0.05);
%Option Theil-Sen
[r1,pval1]=corr(x,y,'Type','Kendall');
[s1, b1] = tsreg(x,y);
S1 =[s1 b1];
y_Theil1=polyval(S1,x);
% Confidence intervals
CI =

the cyclist on 8 Jul 2021
Edited: the cyclist on 8 Jul 2021
tsreg is not a MathWorks function. Are you using this code from the File Exchange?
The wikipedia page for the Theil-Sen method briefly discusses how to calculate the confidence interval, using resampling. Perhaps that will be helpful.
Fabian Moreno on 8 Jul 2021
Thank you for the explanation.