curvefitgui 1.2.0

GUI for the scipy.optimize.curve_fit() function

A GUI for scipy's curve_fit() function

curvefitgui is a graphical interface to the non-linear curvefit function scipy.optimise.curve_fit API reference of the scipy.optimize package. Currently, only the Levenberg-Marquard optimizer is supported. The GUI works with PyQt5 or PyQt6.

Installation

You can install the curvefitgui form PyPi:

pip install curvefitgui

The GUI is supported on Python 3.7 and above.
Note: only curvefitgui is installed without any required dependencies. Depending on if you are using pip or conda to manage your environment you should manually install the following additional packages:

• Using pip:

  pip install numpy scipy matplotlib PyQt6
  

• Using conda:

  conda install numpy scipy matplotlib qtpy pyqt
  

Basic usage

A minimum example to use curvefitgui.curve_fit_gui is:

from curvefitgui import curve_fit_gui
import numpy as np

# define a function for fitting
def f(x, a, b):
    '''
    Linear fit
    function: y = ax + b
    a: slope
    b: intercept
    '''
    return a * x + b

# define x and y data as 1 dimensional numpy arrays of equal length
xdata = np.array([1, 2, 3, 4, 5])
ydata = np.array([-3.5, -2.4, -1, 0.5, 1.8])
        
# execute the function
curve_fit_gui(f, xdata, ydata)   

Arguments

popt, pcov = curve_fit_gui(f, xdata, ydata, xerr=None, yerr=None, p0=None, 
                           xlabel='x-axis', ylabel='y-axis', absolute_sigma=False,
                           jac=None, showgui=True, **kwargs)

curve_fit_gui accepts the following arguments:

• f: callable function that defines the fitfunction. The first argument of f should be the independent variable; other arguments (at least one) are considered to be the fitparameters.
• xdata: 1-D numpy array x-coordinates of the data
• ydata: 1-D numpy array y-coordinates of the data

curve_fit_gui accepts the following keyword arguments:

• yerr: 1-D numpy array, optional (default:None) error/uncertainty in y-values used for weighted fit with a relative weight defined as 1/yerr**2
  (for compatibility also the use of the keyword sigma can be used for the same)
• xerr: 1-D numpy array, optional (default:None) error in x-values. For plotting errorbars only and ignored during fitting
• xlabel: string, optional (default:'x-values') x-axis title in the plot
• ylabel: string, optional (default:'y-values') y-axis title in the plot
• p0: array-like, optional initial values for fit parameters, if not specified 1 is used for each parameter
• showgui: boolean, optional (default=True) if True, the gui is shown, otherwise not
• absolute_sigma: boolean, optional see doc-string scipy.optimize.curve_fit()
• jac: callable, optional see doc-string scipy.optimize.curve_fit()
• kwargs: keyword arguments for compatibility (e.g. you can use sigma to specify the error in y)

Returns

• popt: The values of the fitparameters that minimised the squared residuals if a succesful fit was performed, else None.
• pcov: The estimated covariance of popt. (see also: scipy.optimise.curve_fit API reference)

GUI interface

Once the gui is executed the following window is visible. An explanation of the different controls is described below the figure.

GUI controls

1. Data plot: A matplotlib plot that shows the data as solid dots and both y-error and x-error errorbars if provided. A fitted curve as a dashed line is shown if a fit is performed.
2. Residual plot A matplotlib plot that shows the residuals as the difference between the measured and fitted values: residual = ydata - f(xdata, *fitparameters)
3. Model settings: Here you can enter inital values for the fitparameters. By ticking the chcekbox fix you can set a parameter to fixed:e.g. the parameter is not optmised during the fit.
4. Weight settings: If error data on the y-values are passed using the keyword argument yerr you can use the dropdownbox to set how the error data is treated:
   • None: the error data is ignored
   • Relative: Use the error data for a relative weight. Corresponds to setting scipy's curve_fit() function keyword absolute_sigma = False.
   • Standard deviation: Treat the error data as being standard deviations. Corresponds to setting scipy's curve_fit() function keyword absolute_sigma = True.
5. Evaluate: Use this button to compute the model function given the current values of the parameters (set in the model settings panel)
6. Fit: Performs the fit and updates the parameter values.
7. Report: When a fit is performed, the results are shown here. The information on the model is actually the provided docstring of the function f that is passed to the curvefitgui function.
8. Quit: Quits the gui and returns the fitparameters popt and pcov.
9. Toolbar: This is the standard matplotlib toolbar to adjust some plot properties and provides zoom/pan and save options.
10. FitTextbox: This textbox is generated if a valid fit is performed. It can be moved by the mouse to any convenient positions in the plot.
11. Range Selector Activates/deactivates the range-selector. The range-selector allows to select a datarange used for fitting. Only datapoints that are within the two vertical dashed lines are considered during fitting. The lines can be moved using the mouse.