Approximation coefficients differencies when function fit and Fitting Curve are used

There is a problem of approximation when i use function fit. When i use Fitting Curve i really get good approximation coefficients, because start point parametres are given by program when i choose x and y vectors before fit. When i use fit function start points are random. I can not indicate start points by mysefl and i don't know how can i get the same approximation coefficients when i use function fit and when i use Fitting Curve.

[dpb]

fit has optional 'StartPoint' named parameter so you can select the starting point yourself.

Read all the documentation...

[UserAlex]

I don't know what start points i should choose for adequate resolt

[UserAlex]

And function fit use random start points if i did not enter them, cftool interface use their coefficients

[UserAlex]

function [fitresult, gof] = createFit(NOISEP, SetNumberP)
%CREATEFIT(NOISEP,SETNUMBERP)
%  Create a fit.
%
%  Data for 'untitled fit 1' fit:
%      X Input : NOISEP
%      Y Output: SetNumberP
%  Output:
%      fitresult : a fit object representing the fit.
%      gof : structure with goodness-of fit info.
%
%  See also FIT, CFIT, SFIT.
%  Auto-generated by MATLAB on 25-Oct-2019 20:55:17
%% Fit: 'untitled fit 1'.
[xData, yData] = prepareCurveData( NOISEP, SetNumberP );
% Set up fittype and options.
ft = fittype( 'gauss2' );
opts = fitoptions( 'Method', 'NonlinearLeastSquares' );
opts.Display = 'Off';
opts.Lower = [-Inf -Inf 0 -Inf -Inf 0];
opts.StartPoint = [1 0.987772030980821 0.0127833870316064 0.789954196169235 0.967363517944027 0.0205518543316703];
% Fit model to data.
[fitresult, gof] = fit( xData, yData, ft, opts );
% Plot fit with data.
figure( 'Name', 'untitled fit 1' );
h = plot( fitresult, xData, yData );
legend( h, 'SetNumberP vs. NOISEP', 'untitled fit 1', 'Location', 'NorthEast' );
% Label axes
xlabel NOISEP
ylabel SetNumberP
grid on

Answer 1

[Matt J]

What is "Fitting Curve"? Do you mean the Curve Fitting app? Why do you think the fit function uses different starting points than the app? It shouldn't. It is using the fit() function internally, just as you are attempting to do. You can use the drop down menu option File>>Generate Code in the app to see how it invoked the fit() command to do the fit.

[UserAlex]

When i open code start points are entered by program and i don't know how it gets them

[Matt J]

Post the code that you see, so we can all examine it.

[UserAlex]

function [fitresult, gof] = createFit(NOISEP, SetNumberP)
%CREATEFIT(NOISEP,SETNUMBERP)
%  Create a fit.
%
%  Data for 'untitled fit 1' fit:
%      X Input : NOISEP
%      Y Output: SetNumberP
%  Output:
%      fitresult : a fit object representing the fit.
%      gof : structure with goodness-of fit info.
%
%  See also FIT, CFIT, SFIT.
%  Auto-generated by MATLAB on 25-Oct-2019 20:55:17
%% Fit: 'untitled fit 1'.
[xData, yData] = prepareCurveData( NOISEP, SetNumberP );
% Set up fittype and options.
ft = fittype( 'gauss2' );
opts = fitoptions( 'Method', 'NonlinearLeastSquares' );
opts.Display = 'Off';
opts.Lower = [-Inf -Inf 0 -Inf -Inf 0];
opts.StartPoint = [1 0.987772030980821 0.0127833870316064 0.789954196169235 0.967363517944027 0.0205518543316703];
% Fit model to data.
[fitresult, gof] = fit( xData, yData, ft, opts );
% Plot fit with data.
figure( 'Name', 'untitled fit 1' );
h = plot( fitresult, xData, yData );
legend( h, 'SetNumberP vs. NOISEP', 'untitled fit 1', 'Location', 'NorthEast' );
% Label axes
xlabel NOISEP
ylabel SetNumberP
grid on

[Matt J]

Ah well. That's not convenient. I suggest you post your data in a .mat file, so that we can get an idea of what it looks like and how to fit it.