Source code for neurokit2.epochs.epochs_to_array

# -*- coding: utf-8 -*-
import numpy as np




def epochs_to_array(epochs):
    """**Epochs to Array**

    Convert epochs to a numpy array.

    .. note::

        TODO: We would like to make it work with uneven epochs (not the same length).

    Parameters
    ----------
    epochs : dict
        A dict containing one DataFrame per event/trial. Usually obtained via :func:`epochs_create`.

    Returns
    ----------
    array
        An array containing all signals.


    See Also
    ----------
    events_find, events_plot, epochs_create, epochs_plot

    Examples
    ----------
    .. ipython:: python

      import neurokit2 as nk

      # Get data
      signal = nk.signal_simulate(sampling_rate=100)

      # Create epochs
      epochs = nk.epochs_create(signal, events=[400, 430, 460], sampling_rate=100, epochs_end=1)

      # Convert to epochs
      X = nk.epochs_to_array(epochs)

      @savefig p_epochs_to_array1.png scale=100%
      nk.signal_plot(X.T)
      @suppress
      plt.close()

    """
    example_array = epochs[list(epochs.keys())[0]].select_dtypes(include=["number"])
    if example_array.shape[1] == 2:
        array = np.full((example_array.shape[0], len(epochs)), np.nan)
        for i, key in enumerate(epochs):
            array[:, i] = epochs[key].select_dtypes(include=["number"]).drop("Index", axis=1).values[:, 0]
    else:
        array = np.full((example_array.shape[0], example_array.shape[1] - 1, len(epochs)), np.nan)
        for i, key in enumerate(epochs):
            array[:, :, i] = epochs[key].select_dtypes(include=["number"]).drop("Index", axis=1).values

    return array