# # Spectral Coefficient of Variation (SCV) # --------------------------------------- # # As noted above, the range of log-power values in the theta frequency range is lower # compared to other frequencies, while that of 30-100Hz appear to be quite constant # across the entire frequency axis (homoscedasticity). # # To quantify that, we compute the coefficient of variation (standard deviation/mean) as a # normalized estimate of variance. # ################################################################################################### # Calculate SCV freqs, scv = spectral.compute_scv(sig, fs, nperseg=int(fs), noverlap=0) # Plot the SCV plot_scv(freqs, scv) ################################################################################################### # # As shown above, SCV calculated from the entire segment of data is quite noise due to the # single estimate of mean and standard deviation. To overcome this, we can compute a # bootstrap-resampled estimate of SCV, by randomly drawing slices from the non-overlapping # spectrogram and taking their average. # ################################################################################################### # Calculate SCV with the resampling method
# # Next, let's look at computing the spectral coefficient of variation, with # :func:`~.compute_scv`. # # As noted above, the range of log-power values in the theta frequency range is lower # compared to other frequencies, while that of 30-100Hz appear to be quite constant # across the entire frequency axis (homoscedasticity). # # To quantify that, we compute the coefficient of variation (standard deviation/mean) as a # normalized estimate of variance. # ################################################################################################### # Calculate SCV freqs, scv = compute_scv(sig, fs, nperseg=int(fs), noverlap=0) ################################################################################################### # # There is also a plotting function for SCV, :func:`~.plot_scv`. # ################################################################################################### # Plot the SCV plot_scv(freqs, scv) ################################################################################################### # # As shown above, SCV calculated from the entire segment of data is quite noisy due to the # single estimate of mean and standard deviation.