Decomposition

class nidmd.Decomposition(data=None, filenames=None, sampling_time=None)

Representation of a Decomposition.

Constructor

Decomposition.__init__(data=None, filenames=None, sampling_time=None)

Decomposition Constructor.

Parameters

• data (Array-like) – Preprocessed time-series fMRI data. Can be list of Array-like

• filenames (str) – filenames of .mat files containing data. Can be list of str.

• sampling_time (float, optional) – Sampling time of time-series recording.

Yields

• data (Array-like) – Time-series raw data

• X (Array-like) – Time-series data from t:1->T

• Y (Array-like) – Time-series data from t:0->T-1

• atlas (nidmd.Atlas) – Cortical Parcellation atlas used for this decomposition

• eig_val (Array-like) – Eigenvalues of the eigen-decomposition of the Auto-regressive matrix

• eig_vec (Array-like) – Eigenvectors of the eigen-decomposition of the Auto-regressive matrix

• eig_idx (Array-like) – Indices for descending order of the eigen-decomposition of the Auto-regressive matrix

• A (Array-like) – Auto-regressive matrix

• Z (Array-like) – Approximation of the activity versus time for each mode

• df (pd.DataFrame) – Pandas DataFrame containing the following columns: mode, value, intensity, damping_time, period, conjugate, strength_real, strength_imag, activity

Methods

Decomposition.run()	Run Decomposition.
Decomposition.add(data)	Add data to Decomposition.
Decomposition.extract(filename)	Extracts fMRI data from file.
Decomposition.split(data[, normalize])	Split time-series into X: [1->T] and Y:[0->T-1].
Decomposition.normalize(data[, direction, …])	Normalize a matrix
Decomposition.compute_match(other, m)	Get approximated matched modes for match group with self as a reference.
Decomposition.get_decomposition(x, y)	Get dynamic modes by Least Squares optimization of Auto-regressive model.
Decomposition.dmd([normalize])	Returns a dictionary-like object containing Dynamic Mode Decomposition elements.
Decomposition._compute(val, vec, index, time)	Compute Decomposition to fetch DataFrame with all relevant info.

API reference

class nidmd.Decomposition(data=None, filenames=None, sampling_time=None)

Representation of a Decomposition.

add(data)

Add data to Decomposition.

Parameters

data (Array-like) – Time-series data.

Yields

atlas (nidmd.Atlas) – Cortical Parcellation atlas used for this decomposition

Raises

ImportError – If the import fails.

compute_match(other, m)

Get approximated matched modes for match group with self as a reference. Predicts amplification of approximated modes using linear regression.

Parameters

• other (nidmd.Decomposition) – match group

• m (int) – number of modes analyzed for approximation

Returns

• modes (pd.DataFrame) – Pandas DataFrame containing the following columns: mode, value, damping_time, period, conjudate

• x (Array-like) – Vector containing absolute value of top 10 approximated eigenvalues of self (by mode matching to self)

• y (Array-like) – Vector containing absolute value of top 10 real eigenvalues of self

run()

Run Decomposition.

Returns

df – Pandas DataFrame containing the following columns: mode, value, intensity, damping_time, period, conjugate, strength_real, strength_imag, activity

Return type

pd.DataFrame