def calculate_phi(c_avg, c_avg_next):
c_avg_log = np.log(c_avg)
c_avg_next = np.log(c_avg_next)
phi = Entropy.calculate_averages(c_avg_log, axis=0)
phi_next = Entropy.calculate_averages(c_avg_next, axis=0)
return phi, phi_next
def test_collect_patterns_2_stride_2(self):
data = Entropy.read_series(os.path.join(
self.resource_path, 'perm_en/PermEn_simple_original.txt'),
use_threshold=False,
threshold_value=None)
freqs, mapping = self.permEn.collect_pattern_frequency(data, 2, 2)
npt.assert_almost_equal(freqs, np.array([3 / 5, 2 / 5]), decimal=4)
def test_calculate_permen_simple_2_stride_3(self):
data = Entropy.read_series(os.path.join(
self.resource_path, 'perm_en/PermEn_simple_original.txt'),
use_threshold=False,
threshold_value=None)
res = self.permEn.calculate(3, data, 3)
self.assertAlmostEqual(res, 0, places=4, msg='incorrect PermEn')
def calculate(m, seq, r):
c, c_next = Entropy.calculate_similarity(m, seq, r, include_self_check=True)
number_vectors = len(seq)
c_avg = c / number_vectors
c_next_avg = c_next / (number_vectors - 1)
phi, phi_next = ApproximateEntropy.calculate_phi(c_avg, c_next_avg)
return phi - phi_next
def prepare_calculate_windowed(cls,
m,
file_name,
use_threshold,
threshold_value,
window_size=None,
step_size=None,
calculation_type=None,
dev_coef_value=None,
normalize=False,
stride_size=None):
if not cls.report_cls:
raise NotImplementedError(
'Any Entropy should have its own report type')
res_report = cls.report_cls()
res_report.set_file_name(file_name)
res_report.set_dimension(m)
try:
seq_list, average_rr_list, r_val_list, seq_len = Entropy.prepare_windows_calculation(
file_name,
calculation_type=calculation_type,
dev_coef_value=dev_coef_value,
use_threshold=use_threshold,
threshold_value=threshold_value,
window_size=window_size,
step_size=step_size)
en_results = []
for i in range(len(seq_list)):
calc_kwargs = dict(m=m, seq=seq_list[i], stride=stride_size)
en_results.append(cls.calculate(**calc_kwargs))
except (ValueError, AssertionError):
res_report.set_error("Error! For file {}".format(file_name))
return res_report
res_report.set_seq_len(seq_len)
res_report.set_window_size(window_size)
res_report.set_step_size(step_size)
res_report.set_avg_rr(average_rr_list)
res_report.set_result_values(en_results)
res_report.set_stride_value(1 if stride_size is None else stride_size)
if normalize:
res_report.set_normalized_values(
PermutationEntropy.normalize_series(en_results, m))
return res_report
def calculate(m, seq, r):
c_avg, c_avg_next = Entropy.calculate_similarity(
m, seq, r, include_self_check=False)
return SampleEntropy.calculate_psi(c_avg, c_avg_next)