def test_fooof_checks():
"""Test various checks, errors and edge cases in FOOOF."""
xs, ys = gen_power_spectrum([3, 50], [50, 2], [10, 0.5, 2])
tfm = FOOOF()
# Check dimension error
with raises(ValueError):
tfm.fit(xs, np.reshape(ys, [1, len(ys)]))
# Check shape mismatch error
with raises(ValueError):
tfm.fit(xs[:-1], ys)
# Check trim_spectrum range
tfm.fit(xs, ys, [3, 40])
# Check freq of 0 issue
xs, ys = gen_power_spectrum([3, 50], [50, 2], [10, 0.5, 2])
tfm.fit(xs, ys)
assert tfm.freqs[0] != 0
# Check fit, and string report model error (no data / model fit)
tfm = FOOOF()
with raises(ValueError):
tfm.fit()
def test_fooof_report(skip_if_no_mpl):
"""Check that running the top level model method runs."""
tfm = FOOOF()
tfm.report(*gen_power_spectrum([3, 50], [50, 2], [10, 0.5, 2, 20, 0.3, 4]))
assert tfm
def test_fooof_fit_knee():
"""Test FOOOF fit, with a knee."""
bgp = [50, 2, 1]
gauss_params = [10, 0.5, 2, 20, 0.3, 4]
xs, ys = gen_power_spectrum([3, 50], bgp, gauss_params)
tfm = FOOOF(background_mode='knee')
tfm.fit(xs, ys)
# Note: currently, this test has no accuracy checking at all
assert True
def get_tfm():
"""Get a FOOOF object, with a fit power spectrum, for testing."""
freq_range = [3, 50]
bg_params = [50, 2]
gauss_params = [10, 0.5, 2, 20, 0.3, 4]
xs, ys = gen_power_spectrum(freq_range, bg_params, gauss_params)
tfm = FOOOF()
tfm.fit(xs, ys)
return tfm
def test_fooof_fit_nk():
"""Test FOOOF fit, no knee."""
bgp = [50, 2]
gauss_params = [10, 0.5, 2, 20, 0.3, 4]
xs, ys = gen_power_spectrum([3, 50], bgp, gauss_params)
tfm = FOOOF()
tfm.fit(xs, ys)
# Check model results - background parameters
assert np.all(np.isclose(bgp, tfm.background_params_, [0.5, 0.1]))
# Check model results - gaussian parameters
for ii, gauss in enumerate(group_three(gauss_params)):
assert np.all(np.isclose(gauss, tfm._gaussian_params[ii], [1.5, 0.25, 0.5]))
def test_fooof_fit_failure():
"""Test that fit handles a failure."""
# Use a new FOOOF, that is monkey-patched to raise an error
# This mimicks the main fit-failure, without requiring bad data / waiting for it to fail.
tfm = FOOOF()
def raise_runtime_error(*args, **kwargs):
raise RuntimeError('Test-MonkeyPatch')
tfm._fit_peaks = raise_runtime_error
# Run a FOOOF fit - this should raise an error, but continue in try/except
tfm.fit(*gen_power_spectrum([3, 50], [50, 2], [10, 0.5, 2, 20, 0.3, 4]))
# Check after failing out of fit, all results are reset
for result in get_obj_desc()['results']:
cur_res = getattr(tfm, result)
assert cur_res is None or np.all(np.isnan(cur_res))
############################################################################### # Generate a noisy synthetic power spectrum # Set the frequency range to generate the power spectrum f_range = [1, 50] # Set aperiodic background signal parameters, as [offset, slope] bg_params = [20, 2] # Gaussian peak parameters gauss_params = [10, 1.0, 2.5, 20, 0.8, 2, 32, 0.6, 1] # Set the level of noise to generate the power spectrum with nlv = 0.1 # Create a synthetic power spectrum freqs, spectrum = gen_power_spectrum(f_range, bg_params, gauss_params, nlv) ############################################################################### # Fit an (unconstrained) FOOOF model, liable to overfit fm = FOOOF() fm.report(freqs, spectrum) ############################################################################### # # Notice that in the above fit, we are very likely to think that FOOOF has been overzealous in fitting peaks, and is therefore overfitting. # # This is also suggested by the model r-squared, which is suspiciously high, given the amount of noise we expect. # # To reduce this kind of overfitting, we can update the FOOOF parameters.
