def plot_score_per_time_topo()

in neural/visuals.py [0:0]

• 40 lines of code
• 4 McCabe index (conditional complexity)

def plot_score_per_time_topo(scores_per_time,
                             labels,
                             info,
                             sfreq=120,
                             ref=None,
                             path=None,
                             title=None):
    """
        input:
        -- scores_per_time : list of score_per_time arrays of shape (S, T, C)
        -- labels: list of labels of same len as scores
        -- ref: reference score_per_time, as in an upper bound
        """

    n_models = len(scores_per_time)
    n_subjects, n_times, n_channels = scores_per_time[0].shape

    fig, axes = plt.subplots(n_models, 2)

    for idx in range(n_models):

        # current elements
        score_per_time = scores_per_time[idx]
        if ref is not None:
            score_per_time = ref - score_per_time
        label = labels[idx]

        # take mean over subjects
        evo_data = score_per_time.mean(0).T  # (C, T)
        evo = mne.EvokedArray(evo_data, info=info, tmin=-.500)

        # plot time course
        mne.viz.plot_evoked(
            evo,
            spatial_colors=True,
            scalings=dict(mag=1.),
            show=False,
            axes=axes[idx, 0],
            titles='')
        # ax[0].set_ylim(-.01, .11)
        axes[idx, 0].set_xlabel('time')
        axes[idx, 0].set_ylabel('$\\Delta{}r$')
        axes[idx, 0].set_title('Feature %s' % label)

        # plot topo
        vmax = evo_data.mean(1).max()
        im, _ = mne.viz.plot_topomap(
            evo_data.mean(1),
            evo.info,
            cmap='RdBu_r',
            vmin=-vmax,
            vmax=vmax,
            show=False,
            axes=axes[idx, 1])

        plt.colorbar(im, ax=axes[idx, 1])

    plt.tight_layout()

    if path is not None:
        plt.savefig(path / "sensors_feature_importance.pdf")