def test_full_rank():
n, p = 10, 5
X = np.random.randn(n, p)
X_, _ = full_rank(X)
assert_array_almost_equal(X, X_)
X[:, -1] = X[:, :-1].sum(1)
X_, cond = full_rank(X)
assert cond > 1.e10
assert_array_almost_equal(X, X_)
def test_full_rank():
rng = np.random.RandomState(42)
n, p = 10, 5
X = rng.standard_normal(size=(n, p))
X_, _ = full_rank(X)
assert_array_almost_equal(X, X_)
X[:, -1] = X[:, :-1].sum(1)
X_, cond = full_rank(X)
assert cond > 1.e10
assert_array_almost_equal(X, X_)
def make_first_level_design_matrix(frame_times,
events=None,
hrf_model='glover',
drift_model='cosine',
high_pass=.01,
drift_order=1,
fir_delays=[0],
add_regs=None,
add_reg_names=None,
min_onset=-24,
oversampling=50):
"""Generate a design matrix from the input parameters
Parameters
----------
frame_times : array of shape (n_frames,)
The timing of acquisition of the scans in seconds.
events : DataFrame instance, optional
Events data that describes the experimental paradigm.
The DataFrame instance might have these keys:
'onset': column to specify the start time of each events in
seconds. An error is raised if this key is missing.
'trial_type': column to specify per-event experimental conditions
identifier. If missing each event are labelled
'dummy' and considered to form a unique condition.
'duration': column to specify the duration of each events in
seconds. If missing the duration of each events is set
to zero.
'modulation': column to specify the amplitude of each
events. If missing the default is set to
ones(n_events).
An experimental paradigm is valid if it has an 'onset' key
and a 'duration' key.
If these keys are missing an error will be raised.
For the others keys a warning will be displayed.
Particular attention should be given to the 'trial_type' key
which defines the different conditions in the experimental paradigm.
hrf_model : {'spm', 'spm + derivative', 'spm + derivative + dispersion',
'glover', 'glover + derivative', 'glover + derivative + dispersion',
'fir', None}, optional,
Specifies the hemodynamic response function
drift_model : {'polynomial', 'cosine', None}, optional
Specifies the desired drift model,
high_pass : float, optional
High-pass frequency in case of a cosine model (in Hz).
drift_order : int, optional
Order of the drift model (in case it is polynomial).
fir_delays : array of shape(n_onsets) or list, optional,
In case of FIR design, yields the array of delays used in the FIR
model (in scans).
add_regs : array of shape(n_frames, n_add_reg) or pandas DataFrame, optional
additional user-supplied regressors, e.g. data driven noise regressors
or seed based regressors.
add_reg_names : list of (n_add_reg,) strings, optional
If None, while add_regs was provided, these will be termed
'reg_%i', i = 0..n_add_reg - 1
If add_regs is a DataFrame, the corresponding column names are used
and add_reg_names is ignored.
min_onset : float, optional
Minimal onset relative to frame_times[0] (in seconds)
events that start before frame_times[0] + min_onset are not considered.
oversampling: int, optional,
Oversampling factor used in temporal convolutions.
Returns
-------
design_matrix : DataFrame instance,
holding the computed design matrix, the index being the frames_times
and each column a regressor.
"""
# check arguments
# check that additional regressor specification is correct
n_add_regs = 0
if add_regs is not None:
if isinstance(add_regs, pd.DataFrame):
add_regs_ = add_regs.values
add_reg_names = add_regs.columns.tolist()
else:
add_regs_ = np.atleast_2d(add_regs)
n_add_regs = add_regs_.shape[1]
assert add_regs_.shape[0] == np.size(frame_times), ValueError(
'Incorrect specification of additional regressors: '
'length of regressors provided: %d, number of '
'time-frames: %d' % (add_regs_.shape[0], np.size(frame_times)))
# check that additional regressor names are well specified
if add_reg_names is None:
add_reg_names = ['reg%d' % k for k in range(n_add_regs)]
elif len(add_reg_names) != n_add_regs:
raise ValueError(
'Incorrect number of additional regressor names was provided'
'(%d provided, %d expected' % (len(add_reg_names), n_add_regs))
# computation of the matrix
names = []
matrix = None
# step 1: events-related regressors
if events is not None:
# create the condition-related regressors
if isinstance(hrf_model, str):
hrf_model = hrf_model.lower()
matrix, names = _convolve_regressors(events, hrf_model, frame_times,
fir_delays, min_onset,
oversampling)
# step 2: additional regressors
if add_regs is not None:
# add user-supplied regressors and corresponding names
if matrix is not None:
matrix = np.hstack((matrix, add_regs))
else:
matrix = add_regs
names += add_reg_names
# step 3: drifts
drift, dnames = _make_drift(drift_model, frame_times, drift_order,
high_pass)
if matrix is not None:
matrix = np.hstack((matrix, drift))
else:
matrix = drift
names += dnames
# check column names are all unique
if len(np.unique(names)) != len(names):
raise ValueError('Design matrix columns do not have unique names')
# step 4: Force the design matrix to be full rank at working precision
matrix, _ = full_rank(matrix)
design_matrix = pd.DataFrame(matrix, columns=names, index=frame_times)
return design_matrix