OpenML dataset with id 43884

Author name not available (Why is that?)

Full work available at URL: https://api.openml.org/data/v1/download/22102790/mne-sample-meg-auditory.arff

Upload date: 23 May 2022

Dataset Characteristics

Number of features: 7,668 (numeric: 7,668, symbolic: 0 and in total binary: 0 )
Number of instances: 305
Number of instances with missing values: 0
Number of missing values: 0

Description

MEG data from auditory stimulation experiment using 305 sensors. The design matrix/forward operator is `data[:, :7498]`. The measurements for left stimulation are `data[:, 7498:7583]`. The measurements for right stimulation are `data[:, 7583:]`.

The data was generated with the following script: ``` import mne import numpy as np from mne.datasets import sample from mne.inverse_sparse.mxne_inverse import _prepare_gain

this script used MNE 1.0.2

condition = "Right Auditory" data_path = sample.data_path() + '/MEG/sample' loose = 0 depth = 0.8

fwd_fname = data_path + '/sample_audvis-meg-eeg-oct-6-fwd.fif' ave_fname = data_path + '/sample_audvis-ave.fif' cov_fname = data_path + '/sample_audvis-shrunk-cov.fif'

Read noise covariance matrix

noise_cov = mne.read_cov(cov_fname)

Handling forward solution

forward = mne.read_forward_solution(fwd_fname) targets = {}

for condition in ["Left Auditory", "Right Auditory"]:

   evoked = mne.read_evokeds(
       ave_fname, condition=condition, baseline=(None, 0))
   evoked.crop(tmin=0.04, tmax=0.18)
   evoked = evoked.pick_types(eeg=False, meg=True)

   # Handle depth weighting and whitening (here is no weights)
   forward, gain, gain_info, whitener, _, _ = _prepare_gain(
       forward, evoked.info, noise_cov, pca=False, depth=depth,
       loose=loose, weights=None, weights_min=None, rank=None)

   # Select channels of interest
   sel = [evoked.ch_names.index(name) for name in gain_info['ch_names']]
   M = evoked.data[sel]

   # Whiten data
   M = whitener @ M
   targets[condition] = M

gain is independent of condition:

data = np.hstack([gain, targets["Left Auditory"], targets["Right Auditory"]]) ```

This page was built for dataset: mne-sample-meg-auditory