rafaelscsilva
commented
1 year ago Thank you @PatriciaBota for your precious contribution to BioSPPy that allows automatic feature extraction adapted to biosignals. After carefully reviewing and formatting the code, we can now launch this so-awaited sub-package!
The additions to biosppy made with this PR include:
- Extraction of time domain features (features/time.py): statistical and Hjorth features of the signal and its derivatives
- Extraction of frequency domain features (features/frequency.py): statistical features of the signal's spectrum, power of frequency bands and other spectral-related features
- Extraction of time-frequency domain features (_features/timefreq.py): statistical features of the signal's wavelet decomposition coefficients
- Extraction of phase-space domain features (_features/phasespace.py): recurrence plot features of the signal
- Extraction of cepstral domain features (features/cepstral.py): statistical features of the signal's mel coefficients
- New metrics in the signal_stats function (signals/tools.py): range, q1, q3 and root-mean-square of the signal
- New functions in the stats module (stats.py): histogram, quartiles and diff_stats
Notes:
- The commits specific to EDA signals (signals/eda.py) and EDA plotting (plotting.py) were separated from this PR and implemented in PR#15
- Given the extension of this PR, unit testing for the features sub-package will be considered after merging this PR with the main branch
Added a feature extraction module for different feature domain characteristics: time, frequency, time_frequency, quefrency and phase space.