MATLAB offers a plethora of methods to perform data fitting, based on different minimizers whith different syntaxes and capabilities. The most general one, fmincon, give the user much flexibility in terms of linear constraints and minimization algorithm, at the expense of ease of use.
Fit is a MATLAB class that exposes an user-friendly interface to perfor data fitting operations (or function minimization) using fmincon in a simple, straightforward way.
Only few lines of code are needed to perform a simple data fit:
% Generate data
x = linspace(0, 10, 100);
y = sin(3*x) + 4;
model = @(x, par) sin(par(1)*x) + par(2);
% Initialize the fitting
F = Fit();
F.setData(x, y);
F.setModel(model, 2);
F.setStart([2.8, 4.5]);
% Fit and retrieve parameters!
F.fit()
pars = F.getFittedParameters();while more advanced options can be added easily via other user-friendly methods:
...
fit = Fit();
fit.setData(xData, yData);
fit.setWeights(1./yData); % Weights on data points
fit.setModel(model, 8);
fit.setIRF(IRF); % Fit the data taking automagically into account the system IRF
fit.setStart(fitCond(1,:));
fit.setLb(fitCond(2,:)); % Lower/upper bound support
fit.setUb(fitCond(3,:));
[yFit, xFit] = fit.fitEval(xData); % Evaluate the best fitting model
fitParams = fit.getFittedParameters();
chi2 = fit.getChiSquare()/numel(time); % Get the chi square
parErrors = fit.getParametersErrors(); % Retrieve the parameters uncertainties
...At the moment, the best way to get started is to download/clone the repo, add it to your MATLAB path and run the examples under the folder examples (surprise, surprise).
In the same folder look for the README for a descritpion of the example content.
To check that everythig is supposedly working as expected, try to run the test suite: execute runTest.m, it should run without any error ("Assertion failed").