ridge_results = {}
# store the individual time averaged Wavelet spectra
df_fouriers = pd.DataFrame(index=wAn.periods)
for ID in signals:
wAn.compute_spectrum(signals[ID], do_plot=False)
rd = wAn.get_maxRidge(smoothing_wsize=11)
ridge_results[ID] = rd
df_fouriers[ID] = wAn.get_averaged_spectrum()
# the time-averaged power distribution,
# index holds the signal ids if given, otherwise just a column index
powers_series = em.average_power_distribution(ridge_results.values(),
signal_ids=ridge_results.keys())
# is bi-modal!
pl.power_distribution(powers_series)
# filter out the pure noise signals with a power threshold
high_power_ids = powers_series[powers_series > 10].index
high_power_ridge_results = [ridge_results[ID] for ID in high_power_ids]
# creates a tuple of 4 DataFrames, one summary statistic over time
# for period, amplitude, power and phase each
res = em.get_ensemble_dynamics(high_power_ridge_results)
pl.ensemble_dynamics(*res)
# the Fourier power distribution with pure noise signals
def run_batch(self):
'''
Retrieve all batch settings and loop over the signals
present in the parentDV
'''
dataset_name = self.parentDV.df.name
if self.export_options.isChecked():
OutPath = self.get_OutPath()
if OutPath is None:
return
# TODO: parallelize
ridge_results, df_fouriers = self.do_the_loop()
# check for empty ridge_results
if not ridge_results:
self.NoResults = MessageWindow(
'All ridges below threshold.. no results!', 'No results')
return
# --- compute the time-averaged powers ---
if self.cb_power_dis.isChecked() or self.cb_sorted_powers.isChecked():
powers_series = em.average_power_distribution(
ridge_results.values(), ridge_results.keys(), exclude_coi=True)
if self.cb_power_dis.isChecked():
# plot the distribution
self.pdw = PowerDistributionWindow(powers_series,
dataset_name=dataset_name)
# save out the sorted average powers
if self.export_options.isChecked() and self.cb_sorted_powers.isChecked(
):
fname = os.path.join(OutPath, f'{dataset_name}_ridge-powers.csv')
powers_series.to_csv(fname, sep=',', index=True, header=False)
# --- compute summary statistics over time ---
if self.cb_plot_ens_dynamics.isChecked(
) or self.cb_save_ensemble_dynamics.isChecked():
# res is a tuple of DataFrames, one each for
# periods, amplitude, power and phase
res = em.get_ensemble_dynamics(ridge_results.values())
if self.cb_plot_ens_dynamics.isChecked():
self.edw = EnsembleDynamicsWindow(
res,
dt=self.parentDV.dt,
time_unit=self.parentDV.time_unit,
dataset_name=dataset_name)
if self.export_options.isChecked(
) and self.cb_save_ensemble_dynamics.isChecked():
# create time axis, all DataFrames have same number of rows
tvec = np.arange(res[0].shape[0]) * self.parentDV.dt
for obs, df in zip(['periods', 'amplitudes', 'powers', 'phasesR'],
res):
fname = os.path.join(OutPath, f'{dataset_name}_{obs}.csv')
df.index = tvec
df.index.name = 'time'
df.to_csv(fname, sep=',', float_format='%.3f')
# --- Fourier Distribution Outputs ---
if self.cb_plot_Fourier_dis.isChecked():
self.fdw = FourierDistributionWindow(df_fouriers,
self.parentDV.time_unit,
dataset_name)
if self.export_options.isChecked(
) and self.cb_save_Fourier_dis.isChecked():
fname = os.path.join(OutPath,
f'{dataset_name}_fourier-distribution.csv')
# save out median and quartiles of Fourier powers
df_fdis = pd.DataFrame(index=df_fouriers.index)
df_fdis['Median'] = df_fouriers.median(axis=1)
df_fdis['Q1'] = df_fouriers.quantile(q=0.25, axis=1)
df_fdis['Q3'] = df_fouriers.quantile(q=0.75, axis=1)
df_fdis.to_csv(fname, sep=',', float_format='%.3f')
if self.debug:
print(list(ridge_results.items())[:2])
def run_batch(self):
'''
Retrieve all batch settings and loop over the signals
present in the parentDV
'''
dataset_name = self.parentDV.df.name
if self.export_options.isChecked():
OutPath = self.get_OutPath()
if OutPath is None:
return
# is a dictionary holding the ridge-data
# for each signal and the signal_id as key
ridge_results = self.do_the_loop()
# compute the time-averaged powers
if self.cb_power_dis.isChecked() or self.cb_sorted_powers.isChecked():
powers = em.average_power_distribution(ridge_results.values())
powers_series = pd.Series(index=ridge_results.keys(), data=powers)
# sort by power, descending
powers_series.sort_values(ascending=False, inplace=True)
if self.cb_power_dis.isChecked():
# plot the distribution
self.pdw = PowerDistributionWindow(powers_series,
dataset_name=dataset_name)
# save out the sorted average powers
if self.export_options.isChecked() and self.cb_sorted_powers.isChecked(
):
fname = f'{OutPath}/average_powers_{dataset_name}.csv'
powers_series.to_csv(fname, sep=',', index=True, header=False)
# compute summary statistics over time
if self.cb_plot_ens_dynamics.isChecked(
) or self.cb_save_ensemble_dynamics.isChecked():
# res is a tuple of DataFrames, one each for
# periods, amplitude, power and phase
res = em.get_ensemble_dynamics(ridge_results.values())
if self.cb_plot_ens_dynamics.isChecked():
self.edw = EnsembleDynamicsWindow(
res,
dt=self.parentDV.dt,
time_unit=self.parentDV.time_unit,
dataset_name=dataset_name)
if self.export_options.isChecked(
) and self.cb_save_ensemble_dynamics.isChecked():
# create time axis, all DataFrames have same number of rows
tvec = np.arange(res[0].shape[0]) * self.parentDV.dt
for obs, df in zip(['periods', 'amplitudes', 'powers', 'phasesR'],
res):
fname = f'{OutPath}/{obs}_{dataset_name}.csv'
df.index = tvec
df.index.name = 'time'
df.to_csv(fname, sep=',', float_format='%.3f')
if self.debug:
print(list(ridge_results.items())[:2])
def run_batch(self):
"""
Retrieve all batch settings and loop over the signals
present in the parentDV
"""
dataset_name = self.parentDV.df.name
if self.export_options.isChecked():
OutPath = self.get_OutPath()
if OutPath is None:
return
# TODO: parallelize
ridge_results, df_fouriers = self.do_the_loop()
# check for empty ridge_results
if not ridge_results:
msgBox = QMessageBox()
msgBox.setWindowTitle("No Results")
msgBox.setText("All ridges below threshold.. no results!")
msgBox.exec()
return
settings = QSettings()
float_format = settings.value('float_format', '%.3f')
# --- compute the time-averaged powers ---
if self.cb_power_dis.isChecked() or self.cb_sorted_powers.isChecked():
powers_series = em.average_power_distribution(
ridge_results.values(), ridge_results.keys(), exclude_coi=True)
if self.cb_power_dis.isChecked():
# plot the distribution
self.pdw = PowerDistributionWindow(powers_series,
dataset_name=dataset_name)
# save out the sorted average powers
if self.export_options.isChecked() and self.cb_sorted_powers.isChecked(
):
fname = os.path.join(OutPath, f"{dataset_name}_ridge-powers.csv")
powers_series.to_csv(fname,
sep=",",
float_format=float_format,
index=True,
header=False)
# --- compute summary statistics over time ---
if (self.cb_plot_ens_dynamics.isChecked()
or self.cb_save_ensemble_dynamics.isChecked()):
# res is a tuple of DataFrames, one each for
# periods, amplitude, power and phase
res = em.get_ensemble_dynamics(ridge_results.values())
if self.cb_plot_ens_dynamics.isChecked():
self.edw = EnsembleDynamicsWindow(
res,
dt=self.parentDV.dt,
time_unit=self.parentDV.time_unit,
dataset_name=dataset_name,
)
if (self.export_options.isChecked()
and self.cb_save_ensemble_dynamics.isChecked()):
# create time axis, all DataFrames have same number of rows
tvec = np.arange(res[0].shape[0]) * self.parentDV.dt
for obs, df in zip(["periods", "amplitudes", "powers", "phasesR"],
res):
fname = os.path.join(OutPath, f"{dataset_name}_{obs}.csv")
df.index = tvec
df.index.name = "time"
df.to_csv(fname, sep=",", float_format=float_format)
# --- Fourier Distribution Outputs ---
if self.cb_plot_Fourier_dis.isChecked():
self.fdw = FourierDistributionWindow(df_fouriers,
self.parentDV.time_unit,
dataset_name)
if self.export_options.isChecked(
) and self.cb_save_Fourier_dis.isChecked():
fname = os.path.join(OutPath,
f"{dataset_name}_fourier-distribution.csv")
# save out median and quartiles of Fourier powers
df_fdis = pd.DataFrame(index=df_fouriers.index)
df_fdis["Median"] = df_fouriers.median(axis=1)
df_fdis["Mean"] = df_fouriers.mean(axis=1)
df_fdis["Q1"] = df_fouriers.quantile(q=0.25, axis=1)
df_fdis["Q3"] = df_fouriers.quantile(q=0.75, axis=1)
df_fdis.to_csv(fname, sep=",", float_format=float_format)
if self.debug:
print(list(ridge_results.items())[:2])
Nsignals = len(self.parentDV.df.columns)
msg = f"Processed {Nsignals} signals!"
msgBox = QMessageBox()
msgBox.setWindowTitle("Batch processing done")
msgBox.setText(msg)
msgBox.exec()