skip to navigation
skip to content

Not Logged In

scipy-data_fitting 0.2.4

Data fitting system with SciPy.

Package Documentation

Check out the example fits on Fitalyzer. See the Fitalyzer README for details on how to use Fitalyzer for visualizing your fits.

Documentation

Documentation generated from source with pdoc for the latest version is hosted at packages.python.org/scipy-data_fitting/.

To get started quickly, check out the examples.

Then, refer to the source documentation for details on how to use each class.

Basic usage

from scipy_data_fitting import Data, Model, Fit, Plot

# Load data from a CSV file.
data = Data('linear')
data.path = 'linear.csv'
data.error = (0.5, None)

# Create a linear model.
model = Model('linear')
model.add_symbols('t', 'v', 'x_0')
t, v, x_0 = model.get_symbols('t', 'v', 'x_0')
model.expressions['line'] = v * t + x_0

# Create the fit using the data and model.
fit = Fit('linear', data=data, model=model)
fit.expression = 'line'
fit.independent = {'symbol': 't', 'name': 'Time', 'units': 's'}
fit.dependent = {'name': 'Distance', 'units': 'm'}
fit.parameters = [
    {'symbol': 'v', 'guess': 1, 'units': 'm/s'},
    {'symbol': 'x_0', 'value': 1, 'units': 'm'},
]

# Save the fit result to a json file.
fit.to_json(fit.name + '.json', meta=fit.metadata)

# Save a plot of the fit to an image file.
plot = Plot(fit)
plot.save(fit.name + '.svg')
plot.close()

Controlling the fitting process

The above example will fit the line using the default algorithm `scipy.optimize.curve_fit <http://docs.scipy.org/doc/scipy/reference/generated/scipy.optimize.curve_fit.html>`__.

For a linear fit, it may be more desirable to use a more efficient algorithm.

For example, to use `numpy.polyfit <http://docs.scipy.org/doc/numpy/reference/generated/numpy.polyfit.html>`__, one could set a fit_function and allow both parameters to vary,

fit.parameters = [
    {'symbol': 'v', 'guess': 1, 'units': 'm/s'},
    {'symbol': 'x_0', 'guess': 1, 'units': 'm'},
]
fit.options['fit_function'] = lambda f, x, y, p0, **op: (numpy.polyfit(x, y, 1), )

Controlling the fitting process this way allows, for example, incorporating error values and computing and returning goodness of fit information.

See `scipy_data_fitting.Fit.options <http://packages.python.org/scipy-data_fitting/#scipy_data_fitting.Fit.options>`__ for further details on how to control the fit and also how to use lmfit.

Installation

This package is registered on the Python Package Index (PyPI) at pypi.python.org/pypi/scipy-data_fitting.

Add this line to your application’s requirements.txt:

scipy-data_fitting

And then execute:

$ pip install -r requirements.txt

Or install it yourself as:

$ pip install scipy-data_fitting

Depending on your system configuration, you may need to run the above commands with sudo. Alternatively, you may want to use a virtualenv, which is beyond the scope of this documentation.

Note that the large scientific packages such as NumPy, SciPy, and matplotlib may also be available via your system’s package manager.

To live on the bleeding edge, instead of the package name scipy-data_fitting, you can use this repository directly with

git+https://github.com/razor-x/scipy-data_fitting.git@master#egg=scipy-data_fitting

Note about dependency versions

This package intentionally does not specify dependency versions. Thus, pip will use whatever required packages are currently installed or fetch the latest available version for missing dependencies.

If you want to control what package versions are used, you should specify them explicitly in your project’s own requirements.txt.

Development

Source Repository

The source is hosted at GitHub. Fork it on GitHub, or clone the project with

$ git clone https://github.com/razor-x/scipy-data_fitting.git

Install dependencies with

$ pip install -r requirements.txt

and install the package in development mode with

$ python setup.py develop

Depending on your system configuration, you may need to run the above command with sudo or use a virtualenv.

Note that the large scientific packages such as NumPy, SciPy, and matplotlib may also be available via your system’s package manager.

Documentation

Generate documentation with pdoc by running

$ make docs

Tests

Run the tests with

$ make tests

Examples

Run an example with

$ python examples/example_fit.py

or run all the examples with

$ make examples

License

This code is licensed under the MIT license.

Warranty

This software is provided “as is” and without any express or implied warranties, including, without limitation, the implied warranties of merchantibility and fitness for a particular purpose.

 
File Type Py Version Uploaded on Size
scipy-data_fitting-0.2.4.tar.gz (md5) Source 2014-04-06 22KB
scipy_data_fitting-0.2.4-py3.4.egg (md5) Python Egg 3.4 2014-04-06 43KB
  • Downloads (All Versions):
  • 46 downloads in the last day
  • 468 downloads in the last week
  • 1838 downloads in the last month