Computing circadian rhythmic patterns and beyond: introduction to a new non-Fourier analysis (Q964656)

The authors consider data on the German cockroach locomotor activities which contain records of time moments when the insects blocked one of the infrared sensors in the motion detector boxes. A hierarchical algorithm is proposed for segmentation of these data to the periods of ``activity'' and ``non-activity''. Linear regression is performed with the obtained waveforms to derive the phase and period of the circadian rhythms. The phase response curve is derived using a heterogeneous kernel smoothing algorithm. This approach is compared to the spectral analysis technique on a real data example.