Export all fit results from my curve fitting toolbox for multiple fits

Hi,

I have a fit session using the cftool - there are multiple fits for different sets of data. Some of the data has 2 or 3 different fits.

My ultimate goal is to compare between the different fits for each set of data point (Say for a particular mesaurment I did 3 different fits and I want to get reduced Chi-squared test in order to chose the best fit).

For that I would want to export all of my fits into the workspace as cfit objects. (I don't need any more data, just the cfit files will suffice).

I can manually use the cftool option and extract every one of them wth Fit>Save to workspace and chose only the checkbox with 'Save fit to MATLAB object named:'

However it seems like a redundent manual work, is there a way to do it automatically.

Here's an example of my fits : each row is a unique data set and for some of them I have fit#1,2 and 3 , these are the different versions.

Answer 1

[TED MOSBY]

Hi Pavel,

I referred to the MATLAB documentation for the ‘Curve Fitter Tool’ but couldn’t find a way to automate the export process for every fit result which is generated.

Have a look at the documentation for the Curve Fitter tool here:

www.mathworks.com/help/curvefit/generating-code-and-exporting-fits-to-the-workspace.html

I can suggest an alternative way to analyse the fits exported but for that you need to first export the fits manually to the workspace. Post that you can write a script that automatically analyse the fits and calculate the reduced chi-squared. A dummy example is shown below:

% Assuming fitresult_lin and fitresult_quad are in the workspace 
% And assuming x, y1, and y2 are your data vectors 
% Number of parameters for each model: 
% For a linear fit (ax + b), m_lin = 2 
% For a quadratic fit (ax^2 + bx + c), m_quad = 3 
m_lin = 2; 
m_quad = 3; 

% Calculate residuals for each fit 
residuals_lin = y1 - feval(fitresult_lin, x); 
residuals_quad = y2 - feval(fitresult_quad, x); 

% Assuming equal variance (sigma = 1) for simplicity 
sigma = 1; 

% Calculate reduced Chi-squared for each fit 
chi_squared_red_lin = sum((residuals_lin / sigma).^2) / (length(y1) - m_lin); 
chi_squared_red_quad = sum((residuals_quad / sigma).^2) / (length(y2) - m_quad); 

% Display the reduced Chi-squared values 
fprintf('Reduced Chi-squared (Linear Fit): %f\n', chi_squared_red_lin); 
fprintf('Reduced Chi-squared (Quadratic Fit): %f\n', chi_squared_red_quad); 

% Comparing the reduced Chi-squared values 
if chi_squared_red_lin < chi_squared_red_quad 
    fprintf('The linear fit is better according to the reduced Chi-squared criterion.\n'); 
else 
    fprintf('The quadratic fit is better according to the reduced Chi-squared criterion.\n'); 
end 

As this is just a dummy example, feel free to adjust it according to your data and fits.

Hope this helps!