betterfit 0.0.2a3

Regression of data with error propagation

Betterfit

Installation

pip install betterfit

Getting Started

Refer to getting_started.py:

# getting_started.py
import matplotlib.pyplot as plt
import sympy as smp
from uncertainties import ufloat

from betterfit import Dataset, LinearFit

# create immutable datasets
resistance_data = Dataset.fromiter(symbol='R',
                                   values=[2000, 1800, 1600, 1400, 1200, 1000, 900],
                                   errors=10)
length_data = Dataset.fromiter(symbol=smp.Symbol('l'),
                               values=[160, 172, 192, 223, 254, 298, 327],
                               errors=[1.5, 2.4, 2, 2.3, 2.1, 3, 5])
length_data = length_data.multiply(1e-3)

# define quantities as sympy symbols
R, l, d = smp.symbols('R l d')

# define an expression depending on these symbols
phi = smp.atan(l / d)

# perform linear fit
linearfit = LinearFit()
linearfit.add_constant(symbol=d,
                       value=ufloat(0.535, 0.01))
linearfit.add_datasets(resistance_data, length_data)
slope, yintercept = linearfit.fit(x_expr=1 / phi,
                                  y_expr=R)

# linearfit.fit return sympy symbols representing the slope and yintercept
# the values can be accessed through linearfit[slope] and linearfit[yintercept]
print(f'slope: {linearfit[slope]:.3g}')
print(f'yintercept: {linearfit[yintercept]:.3g}')

# similarily, values of expressions can be accessed through linearfit[expr]
print(f'phi: {linearfit[phi]}')

# plot setup
fig, ax = plt.subplots()
ax.set_title('Example Linear Fit with Weighted Least Squares')
ax.set_xlabel(r'$\frac{1}{\phi}$ [$s$]')
ax.set_ylabel(r'$R$ [$\Omega$]')

# plot fit
linearfit.plot_on(ax,
                  x=1 / phi,
                  y=R,
                  fmt='o',
                  label='data')
linearfit.plot_fit_on(ax)

ax.legend()

plt.show()