.. DO NOT EDIT.
.. THIS FILE WAS AUTOMATICALLY GENERATED BY SPHINX-GALLERY.
.. TO MAKE CHANGES, EDIT THE SOURCE PYTHON FILE:
.. "auto_examples/00_encodingdecoding/linear_encoding_model.py"
.. LINE NUMBERS ARE GIVEN BELOW.

.. only:: html

    .. note::
        :class: sphx-glr-download-link-note

        :ref:`Go to the end <sphx_glr_download_auto_examples_00_encodingdecoding_linear_encoding_model.py>`
        to download the full example code

.. rst-class:: sphx-glr-example-title

.. _sphx_glr_auto_examples_00_encodingdecoding_linear_encoding_model.py:


Linear encoding model
=============================================

When it comes to fitting encoding models to fMRI data,
the most common approach is to use a linear encoding model,
were the predicted BOLD response is a linear combination of
different neural populations with predefined tuning properties.

Here we explore such an approach.

.. GENERATED FROM PYTHON SOURCE LINES 13-34

.. code-block:: Python


    # Import necessary libraries
    from braincoder.models import VonMisesPRF
    import numpy as np
    import pandas as pd
    import matplotlib.pyplot as plt

    # Set up six evenly spaced von Mises PRFs
    centers = np.linspace(0.0, 2*np.pi, 6, endpoint=False)
    parameters = pd.DataFrame({'mu':centers, 'kappa':1., 'amplitude':1.0, 'baseline':0.0},
                              index=pd.Index([f'Voxel {i+1}' for i in range(6)], name='voxel'))

    # We have 3 voxels, each with a linear combination of the 6 von Mises functions:
    weights = np.array([[1, 0, 1],
                        [1, .5, 1],
                        [0, 1, 0],
                        [0, .5, 0],
                        [0, 0, 1],
                        [0, 0, 1]]).astype(np.float32)

    model = VonMisesPRF(parameters=parameters, weights=weights)







.. GENERATED FROM PYTHON SOURCE LINES 35-44

.. code-block:: Python


    # Plot the basis functions
    # Note that the function `basis_functions` returns a `tensorflow` `Tensor`,
    # which has to be converted to a numpy array:
    orientations = np.linspace(0, np.pi*2, 100)
    basis_responses = model.basis_predictions(orientations, parameters).numpy()

    _ = plt.plot(orientations, basis_responses)




.. image-sg:: /auto_examples/00_encodingdecoding/images/sphx_glr_linear_encoding_model_001.png
   :alt: linear encoding model
   :srcset: /auto_examples/00_encodingdecoding/images/sphx_glr_linear_encoding_model_001.png
   :class: sphx-glr-single-img





.. GENERATED FROM PYTHON SOURCE LINES 45-51

.. code-block:: Python


    # Plot the predicted responses for the 3 voxels
    # Each voxel timeseries is a weighted sum of the six basis functions
    pred = model.predict(paradigm=orientations)
    _ = plt.plot(orientations, pred)




.. image-sg:: /auto_examples/00_encodingdecoding/images/sphx_glr_linear_encoding_model_002.png
   :alt: linear encoding model
   :srcset: /auto_examples/00_encodingdecoding/images/sphx_glr_linear_encoding_model_002.png
   :class: sphx-glr-single-img





.. GENERATED FROM PYTHON SOURCE LINES 52-73

.. code-block:: Python


    # Import the weight fitter
    from braincoder.optimize import WeightFitter
    from braincoder.utils import get_rsq

    # Simulate data
    data = model.simulate(paradigm=orientations, noise=0.1)

    # Fit the weights
    weight_fitter = WeightFitter(model, parameters, data, orientations)
    estimated_weights = weight_fitter.fit(alpha=0.1)

    # Get predictions for the fitted weights
    pred = model.predict(paradigm=orientations, weights=estimated_weights)
    r2 = get_rsq(data, pred)

    # Plot the data and the predictions
    plt.figure()
    plt.plot(orientations, data, c='k')
    plt.plot(orientations, pred.values, c='k', ls='--')
    plt.title(f'R2 = {r2.mean():.2f}')
    # %%


.. image-sg:: /auto_examples/00_encodingdecoding/images/sphx_glr_linear_encoding_model_003.png
   :alt: R2 = 0.69
   :srcset: /auto_examples/00_encodingdecoding/images/sphx_glr_linear_encoding_model_003.png
   :class: sphx-glr-single-img


.. rst-class:: sphx-glr-script-out

 .. code-block:: none


    Text(0.5, 1.0, 'R2 = 0.69')




.. rst-class:: sphx-glr-timing

   **Total running time of the script:** (1 minutes 7.599 seconds)


.. _sphx_glr_download_auto_examples_00_encodingdecoding_linear_encoding_model.py:

.. only:: html

  .. container:: sphx-glr-footer sphx-glr-footer-example

    .. container:: sphx-glr-download sphx-glr-download-jupyter

      :download:`Download Jupyter notebook: linear_encoding_model.ipynb <linear_encoding_model.ipynb>`

    .. container:: sphx-glr-download sphx-glr-download-python

      :download:`Download Python source code: linear_encoding_model.py <linear_encoding_model.py>`


.. only:: html

 .. rst-class:: sphx-glr-signature

    `Gallery generated by Sphinx-Gallery <https://sphinx-gallery.github.io>`_