N_pca_components=40, on the other hand, would actually reduce the Reconstruct the data using all 306 PCA components. Later exclude) only large, dominating artifacts in the data, but still For example, full-rankģ06-channel MEG data might use n_components=40 to find (and This is not typically used for EEG data, but for MEG data, it’s common to The n_channels PCA components the ICA algorithm will actually fit. The n_components parameter determines how many components out of Alternatively, using n_components as a float will also avoid Otherwise numerical instability causes problems when computing the mixing Important to remove any numerically-zero-variance components in the data, Used for dimensionality reduction of the data, or dealing with low-rankĭata (such as those with projections, or MEG data processed by SSS). Reconstructing the data when calling apply(). N_pca_components determines how many PCA components will be kept when Restores any data not passed to the ICA algorithm, i.e., the PCAĬomponents between n_components and n_pca_components. Includes ICA components based on ica.include and ica.exclude. ![]() Unmixes the data with the unmixing_matrix_. Passing the n_components largest-variance components to the ICAĪlgorithm to obtain the unmixing matrix (and by pseudoinversion, the (using noise_cov if provided, or the standard deviation of eachĬhannel type) and then principal component analysis (PCA). Whitening the data by means of a pre-whitening step A warning will be emitted otherwise.Ī trailing _ in an attribute name signifies that the attribute wasĪdded to the object during fitting, consistent with standard scikit-learn High-pass filter, but not baseline correct the data for good If you intend to fit ICA on Epochs, it is recommended to Iterations it took ICA.fit() to complete will be stored in theĪllow ICA on MEG reference channels. Will set maximum iterations to 1000 for 'fastica'Īnd to 500 for 'infomax' or 'picard'. Allowed entries are determined by the various algorithm fit_params dict | NoneĪdditional parameters passed to the ICA estimator as specified by Defaults to 'fastica'.įor reference, see. Infomax, set method='infomax' and fit_params=dict(extended=True) To achieve reproducible results, pass a value here to explicitly initialize Likely produce different output every time this function or method is run. (see RandomState for details), meaning it will most The seed will be obtained from the operating system random_state None | int | instance of RandomStateĪ seed for the NumPy random number generator (RNG). If None (default), channelsĪre scaled to unit variance (“z-standardized”) as a group by channel The default (None) will also take into account the N_components_ (note the trailing underscore).Ĭhanged in version 0.22: For a float, the number of components will accountįor greater than the given variance level instead of less than orĮqual to it. ICA.fit() method will be stored in the attribute The actual number used when executing the Stability problems when whitening, particularly when working withĭefaults to None. Third component, on the other hand, would be excluded.Ġ.999999 will be used. Requested threshold of 80% explained variance can be exceeded. Passing 0.8 here (corresponding to 80% ofĮxplained variance) would yield the first two components,Įxplaining 90% of the variance: only by using both components the The cumulative variance of the data greater than n_components.Ĭonsider this hypothetical example: we have 3 components, the firstĮxplaining 70%, the second 20%, and the third the remaining 10% Will select the smallest number of components required to explain Must be greater than 1 and less than or equal to the number of ![]() Number of principal components (from the pre-whitening PCA step) thatĪre passed to the ICA algorithm during fitting: Parameters : n_components int | float | None ![]() Typically, a cutoff frequency of 1 Hz is recommended. Requires the data to be high-pass filtered prior to fitting. ICA is sensitive to low-frequency drifts and therefore
0 Comments
Leave a Reply. |
Details
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |