def test_surrogate_analysis():
rng = np.random.RandomState(0)
data = np.load("../examples/data/eeg_32chans_10secs.npy")
ts1 = data[0, 0:512].ravel()
ts2 = data[1, 0:512].ravel()
p_vals, surr_vals, surrogates, r_value = surrogate_analysis(
ts1,
ts2,
num_surr=1000,
estimator_func=None,
ts1_no_surr=False,
rng=rng)
expected_result = np.load("data/test_ts_surrogates.npy").squeeze()
np.testing.assert_array_equal(surrogates, expected_result)
expected_result = np.load("data/test_ts_surrogates_p_val.npy").squeeze()
assert p_vals == expected_result
np.save("/tmp/out.npy", surr_vals)
expected_result = np.load("data/test_ts_surrogates_corr.npy").squeeze()
np.testing.assert_array_almost_equal(surr_vals, expected_result)
def test_surrogate_analysis_fdr():
rng = np.random.RandomState(0)
data = np.load("../examples/data/eeg_32chans_10secs.npy")
ts1 = data[0, 0:512].ravel()
ts2 = data[1, 0:512].ravel()
p_val, surr_vals, surrogates, r_value = surrogate_analysis(
ts1,
ts2,
num_surr=1000,
estimator_func=None,
ts1_no_surr=False,
rng=rng)
num_ts = 2
num_pvals = num_ts * (num_ts - 1) / 2.0
num_pvals = np.int32(num_pvals)
p_vals = np.ones([num_pvals, 1]) * p_val
h, crit_p = fdr(p_vals, 0.01, "pdep")
expected_result = np.load("data/test_ts_surrogates_fdr_h.npy")
assert h == expected_result
expected_result = np.load("data/test_ts_surrogates_fdr_crit_p.npy")
assert crit_p == expected_result
def test_surrogate_analysis2():
rng = np.random.RandomState(0)
data = np.load("../examples/data/eeg_32chans_10secs.npy")
ts1 = data[0, 0:512].ravel()
ts2 = data[1, 0:512].ravel()
p_val, surr_vals, surrogates, r_value = surrogate_analysis(
ts1, ts2, num_surr=1000, ts1_no_surr=True, rng=rng)
expected_result = np.load("data/test_ts_surrogates2.npy")
np.testing.assert_array_equal(surrogates, expected_result)
expected_result = np.load("data/test_ts_surrogates2_p_val.npy")
assert p_val == expected_result
# -*- coding: utf-8 -*-
import numpy as np
np.set_printoptions(precision=3, linewidth=256)
from dyconnmap.ts import fdr, surrogate_analysis
if __name__ == "__main__":
rng = np.random.RandomState(0)
data = np.load(
"/home/makism/Github/dyconnmap/examples/data/eeg_32chans_10secs.npy")
ts1 = data[0, :].ravel()
ts2 = data[1, :].ravel()
p_val, corr_surr, surrogates, r_value = surrogate_analysis(
ts1, ts2, num_surr=1000, ts1_no_surr=True, rng=rng)
num_ts = 2
p_vals = np.ones([num_ts * (num_ts - 1) / 2, 1]) * p_val
q = 0.01
method = 'pdep'
h, crit_p = fdr(p_vals, q, method)
print("p-value: {0}, h: {1} (critical p-value: {2})".format(
p_val, h, crit_p))