.. DO NOT EDIT.
.. THIS FILE WAS AUTOMATICALLY GENERATED BY SPHINX-GALLERY.
.. TO MAKE CHANGES, EDIT THE SOURCE PYTHON FILE:
.. "auto_examples/sns/plot_01_sns_basics.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_sns_plot_01_sns_basics.py>`
        to download the full example code.

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

.. _sphx_glr_auto_examples_sns_plot_01_sns_basics.py:


Basic Sensor Noise Suppression
==============================

This example adds independent sensor noise to a spatially redundant signal,
fits SNS, and applies the same workflow to an MNE Raw object.

.. GENERATED FROM PYTHON SOURCE LINES 8-56



.. image-sg:: /auto_examples/sns/images/sphx_glr_plot_01_sns_basics_001.png
   :alt: Sensor Noise Suppression — channel 0
   :srcset: /auto_examples/sns/images/sphx_glr_plot_01_sns_basics_001.png
   :class: sphx-glr-single-img


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

 .. code-block:: none

    Relative error before SNS: 0.201
    Relative error after SNS:  0.117
    Effective neighbors: 12
    Samples rejected while fitting: 0
    RawArray (24, 2500)






|

.. code-block:: Python


    import mne
    import numpy as np

    from mne_denoise.sns import SNS
    from mne_denoise.viz import plot_signal_overlay

    rng = np.random.default_rng(12)
    n_channels, n_sources, n_times = 24, 4, 2500
    sources = rng.standard_normal((n_sources, n_times))
    shared = rng.standard_normal((n_channels, n_sources)) @ sources
    observed = shared + 0.45 * rng.standard_normal(shared.shape)

    model = SNS(n_neighbors=12, outlier_threshold=8.0)
    cleaned = model.fit_transform(observed)

    error_before = np.linalg.norm(observed - shared) / np.linalg.norm(shared)
    error_after = np.linalg.norm(cleaned - shared) / np.linalg.norm(shared)
    print(f"Relative error before SNS: {error_before:.3f}")
    print(f"Relative error after SNS:  {error_after:.3f}")
    print(f"Effective neighbors: {model.n_neighbors_}")
    print(f"Samples rejected while fitting: {model.rejected_sample_count_}")

    plot_signal_overlay(
        observed,
        cleaned,
        np.arange(n_times) / 250.0,
        pick=0,
        start=0.0,
        stop=1.6,
        scale_after=False,
        before_label="Observed",
        after_label="SNS",
        reference=shared[0],
        reference_label="Shared signal",
        x_label="Time (s)",
        title="Sensor Noise Suppression — channel 0",
        show=False,
    )

    # MNE containers use the same estimator. The returned object is a copy with its
    # metadata and channel layout preserved.
    info = mne.create_info(
        [f"EEG {index:03d}" for index in range(n_channels)], 250.0, "eeg"
    )
    raw = mne.io.RawArray(observed, info, verbose=False)
    raw_clean = SNS(n_neighbors=12).fit_transform(raw)
    print(type(raw_clean).__name__, raw_clean.get_data().shape)


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

   **Total running time of the script:** (0 minutes 0.343 seconds)


.. _sphx_glr_download_auto_examples_sns_plot_01_sns_basics.py:

.. only:: html

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

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

      :download:`Download Jupyter notebook: plot_01_sns_basics.ipynb <plot_01_sns_basics.ipynb>`

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

      :download:`Download Python source code: plot_01_sns_basics.py <plot_01_sns_basics.py>`

    .. container:: sphx-glr-download sphx-glr-download-zip

      :download:`Download zipped: plot_01_sns_basics.zip <plot_01_sns_basics.zip>`