symfit 0.3.2

Symbolic Fitting; fitting as it should be.

Documentation: http://symfit.readthedocs.org/

This project aims to marry the power of scipy.optimize with the readability of SymPy to create a highly readable and easy to use fitting package which works for projects of any scale.

symfit is designed to be very readable:

x = variables('x')
A, sig, x0 = parameters('A, sig, x0')

# Gaussian distribution
gaussian = A * exp(-(x - x0)**2 / (2 * sig**2))

fit = Fit(gaussian, xdata, ydata)
fit_result = fit.execute()

You can also name dependent variables, allowing for sexy assignment of data:

x, y = variables('x, y')
model = {y: a * x**2}

fit = Fit(model, x=xdata, y=ydata, sigma_y=sigma)
fit.execute()

Constraint maximization has never been this easy:

x, y = parameters('x, y')
model = 2*x*y + 2*x - x**2 -2*y**2
constraints = [
    Eq(x**3 - y, 0),
    Ge(y - 1, 0),
]

fit = Maximize(model, constraints=constraints)
fit_result = fit.execute()

And evaluating a model with the best fit parameters is easy since symfit expressions are callable:

y = gaussian(x=xdata, **fit_result.params)

For many more features such as bounds on Parameter’s, maximum-likelihood fitting, and much more check the docs at http://symfit.readthedocs.org/.

You can find symfit on github at https://github.com/tBuLi/symfit.