MATLAB: Options when drawing using the Curve Fitting tool

curve fitting toolnonlinear regression

I am inputting x y z data and using 'Curve Fitting tool' to create a nonlinear regression graph.

I am trying to find the coefficients a,b,c,d suitable for the data using the 'Custom Equation' formula, but the graph is too different from the data.

I tried changing Robust and Algorithm, but the graph is not similar to the data.

What additional work should I do to get the graph?

I also attached a data file.

Best Answer

Your datapoints seem to have a polynomial relationship, not logarithmic. I think your should change your model. In addition, visual check is not enough for curve fitting. There are lots of things to check after fitting the model in order to reach a good model. Please read:

https://www.mathworks.com/help/curvefit/curve-fitting.html

Especially fit potsprocessing > residual analysis.

Related Solutions

MATLAB: Options when drawing using the Curve Fitting tool

Hi, post your data and function in txt format, if possible.

MATLAB: Matlab curve fitting tool automation for selection of fit with minimum sse.

You could certainly write your own function by using the gof output of the "fit" function.

Here is an example that shows you how to use the "fit" to see the Goodness of Fit of different methods. You could store the "gof" values in an array and find the max GoF.

function bestFitName = Gof_Finder( x, y )
    Fits = {'poly1', 'poly2', 'poly3', 'exp1'};
      for i=1:numel(Fits)
          [~,GoFs(i)] = fit( x, y,  Fits{i} );
      end
      [~, minIdx] = min([GoFs.sse]);
      bestFitName = Fits{minIdx};
  end

The above function will return the name of the best fit of the data x and y.

Best Answer

Related Solutions

MATLAB: Options when drawing using the Curve Fitting tool

MATLAB: Matlab curve fitting tool automation for selection of fit with minimum sse.

Related Question