Skip to content

Latest commit

 

History

History
143 lines (112 loc) · 4.38 KB

README.md

File metadata and controls

143 lines (112 loc) · 4.38 KB
Raw

FitAPP

Interactively plot 1D data and fit it with custom models from the comfort of your browser window. The main goal is to provide a simple and efficient GUI for the well known function scipy.optimize.curve_fit(). Use cases:

  • An alternative to interactive matplotlib plots in jupyter notebook (with limited functionality).
  • I need to fit these 50 heat flux profiles by hand until lunch. It would really help if I could...
    • quickly adjust the initial parameters when needed
    • discard obvious outliers from the fitted dataset
    • inspect the resulting curve

The frontend is written using the Svelte framework and it communicates with python (to load data, calculate the fit, etc.) via a local Flask server.

Install

Clone the repo to your machine:

git clone -b build https://github.com/hetsko/fitapp.git

The option -b build selects the build branch, which includes (hopefully) the latest version of the javascript build, which is required to run the app.

Alternatively, see section Development for instructions on how to build the frontend yourself.

Usage

Import as any other python package and create a new instance of the app:

from fitapp.server import get_fitapp, Data

# 1. Create the app, register labels (arbitrary names for your data sets).
fit = get_fitapp(port=8282)
fit.labels = ['data1', 'data2']

# 2. A new tab with the app should be launched automatically in your default
#    web browser (http://127.0.0.1:8282)
#

# 3. Register a data callback function that returns your data on demand. It
#    has to return an instance of Data(), which accepts any of the following:
#       - `list`
#       - `numpy.ndarray`
#       - `xarray.DataArray`
@fit.callback_data
def get_data(label):
    if label == 'data1':
        return Data(x=[1, 2, 3], y=[4, 10, -3])
    elif label == 'data1':
        return Data(x=[1, 2, 3], y=[-1, -2, -3])

# 4. Register a fit callback function that is used to fit the data. It is
#    passed directly to scipy.optimize.curve_fit().
#    Note: It may be beneficial to also specify @fit.callback_guess (see below)
@fit.set_fitfunc
def func(x, A, B):
    return A*x + B

Additionally, the app offeres a few more callbacks, which are optional:

# (Optional) Provide a longer description for the datasets. Return `str`.
@fit.callback_metadata
def get_guess(label):
    return {
        'data1': 'Dataset 1, monday, weather=sunny',
        'data2': 'Dataset 2, friday, weather=raining',
    }[label]

# (Optional) Provide default initial values for the fit parameters.
#            Return `list`, `numpy.ndarray`, `xarray.DataArray`.
@fit.callback_guess
def get_guess(label):
    #       A  B
    return [1, 0]

The current fit results are contained in a dict object:

>>> fit.fit_results
{'args': array([-3011.91991276,   943.59548712]),
 'argsErr': array([361.71457033,  86.88364269]),
 'params': array(['A', 'B'], dtype='<U1'),
 'label': (20631, 11),
 'guess': array([1, 1]),
 'data': Data(x, y, yerr),
 'selected': array([22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35])}

The app starts running in a background thread the moment you run get_fitapp() and keeps running until you end the script or close the jupyter notebook. To create more than one app, specify different port for each of them (pick an arbitrary 4-digit number). When the app is running, feel free to change the get_data callback and fitfunc any time by re-applying the appropriate decorators to the new functions and refreshing the browser window.

Mouse & keyboard control of the app

  • MouseWheel - zoom in, zoom out
  • Ctrl + LeftMouse - drag the mouse cursor to zoom in on the selected area
  • Shift + LeftMouse - drag the mouse cursor to select points to fit
  • Ctrl + Shift + LeftMouse - drag the mouse cursor to unselect points to fit
  • Toggle on the "Fit data" switch in the panel on the right to see the fitted curve

Development

Install the dependencies

cd svelte-app

npm install

python -m venv venv  # Requires python 3.7+
pip install -r requirements.txt

Then start Rollup (automatically compiles the svelte app when code is changed)

npm run dev

and the flask server

python

and navigate to 127.0.0.1:5050.

To create an optimized version of the app use

npm run build