skdh.features.DetailPowerRatio

class skdh.features.DetailPowerRatio(wavelet='coif4', freq_band=None)

The ratio of the power in the detail signals that span the specified frequency band. Uses the discrete wavelet transform to break down the signal into constituent components at different frequencies.

Parameters:

waveletstr

Wavelet to use. Options are the discrete wavelets in PyWavelets. Default is ‘coif4’.

freq_bandarray_like

2-element array-like of the frequency band (Hz) to get the power in. Default is [1, 10].

Attributes:

f_band

wave

Methods

compute(signal[, fs, axis])

Compute the detail power ratio

Notes

In the original paper [1], the result is multiplied by 100 to obtain a percentage. This final multiplication is not included in order to obtain results that have a scale that closer matches the typical 0-1 (or -1 to 1) scale for machine learning features. NOTE that this does not mean that the values will be in this range - since the scaling factor is the original acceleration and not the wavelet detail values.

References

[1]

Sekine, M. et al. “Classification of waist-acceleration signals in a continuous walking record.” Medical Engineering & Physics. Vol. 22. Pp 285-291. 2000.

compute(signal, fs=1.0, *, axis=-1)

Compute the detail power ratio

Parameters:

signalarray-like

Array-like containing values to compute the detail power ratio for.

fsfloat, optional

Sampling frequency in Hz. If not provided, default is 1.0Hz.

axisint, optional

Axis along which the signal entropy will be computed. Ignored if signal is a pandas.DataFrame. Default is last (-1).

Returns:

power_rationumpy.ndarray

Computed detail power ratio.