.. DO NOT EDIT.
.. THIS FILE WAS AUTOMATICALLY GENERATED BY SPHINX-GALLERY.
.. TO MAKE CHANGES, EDIT THE SOURCE PYTHON FILE:
.. "auto_examples/ssa/plot_02_local_ssa_properties.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_ssa_plot_02_local_ssa_properties.py>`
        to download the full example code.

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

.. _sphx_glr_auto_examples_ssa_plot_02_local_ssa_properties.py:


Local SSA properties on a synthetic signal
===========================================

This is an algorithm demonstration, not a replication of Teixeira et al.'s
figures. The desired signal is broadband so that the example exercises the
paper's high-energy-structure-versus-residual assumption without constructing
an unrealistically perfect separability case.

.. GENERATED FROM PYTHON SOURCE LINES 10-47



.. image-sg:: /auto_examples/ssa/images/sphx_glr_plot_02_local_ssa_properties_001.png
   :alt: Teixeira local SSA synthetic demonstration
   :srcset: /auto_examples/ssa/images/sphx_glr_plot_02_local_ssa_properties_001.png
   :class: sphx-glr-single-img


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

 .. code-block:: none

    Correlation with broadband reference: 0.975
    Reference gain: 0.941
    MDL dimensions: [2 1 1 1 1 2]

    <Figure size 2400x800 with 1 Axes>





|

.. code-block:: Python


    import numpy as np

    from mne_denoise.ssa import LocalSingularSpectrumAnalysis
    from mne_denoise.viz import plot_signal_overlay

    sfreq = 128.0
    times = np.arange(1000) / sfreq
    rng = np.random.default_rng(8)
    neural = 0.5 * rng.standard_normal(times.size)
    artifact = 6.0 * np.sin(2 * np.pi * 0.5 * times)
    observed = neural + artifact + 0.05 * rng.standard_normal(times.size)

    cleaner = LocalSingularSpectrumAnalysis(
        window_length=41,
        n_clusters=6,
        random_state=0,
    )
    cleaned = cleaner.fit_transform(observed[np.newaxis])[0]
    correlation = np.corrcoef(cleaned, neural)[0, 1]
    gain = np.dot(cleaned, neural) / np.dot(neural, neural)
    print(f"Correlation with broadband reference: {correlation:.3f}")
    print(f"Reference gain: {gain:.3f}")
    print(f"MDL dimensions: {cleaner.subspace_dimensions_[0]}")

    plot_signal_overlay(
        observed,
        cleaned,
        times,
        reference=neural,
        before_label="Observed",
        after_label="Local SSA residual",
        reference_label="Broadband reference",
        scale_after=False,
        title="Teixeira local SSA synthetic demonstration",
        show=False,
    )


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

   **Total running time of the script:** (0 minutes 0.366 seconds)


.. _sphx_glr_download_auto_examples_ssa_plot_02_local_ssa_properties.py:

.. only:: html

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

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

      :download:`Download Jupyter notebook: plot_02_local_ssa_properties.ipynb <plot_02_local_ssa_properties.ipynb>`

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

      :download:`Download Python source code: plot_02_local_ssa_properties.py <plot_02_local_ssa_properties.py>`

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

      :download:`Download zipped: plot_02_local_ssa_properties.zip <plot_02_local_ssa_properties.zip>`