http://arxiv.org/abs/1910.08799

The expert system for time series analysis of irregularly spaced signals is reviewed. It consists of a number of complementary algorithms and programs, which may be effective for different types of variability. Obviously, for a pure sine signal, all the methods should produce the same results. However, for irregularly spaced signals with a complicated structure, e.g. a sum of different components, different methods may produce significantly different results.
The basic approach is based on classical method of the least squares (1994OAP…..7…49A). However, contrary to common “step-by-step” methods of removal important components (e.g. mean, trend (“detrending”), sine wave (“prewhitening”), where covariations between different components are ignored, i.e. erroneusly assumed to be zero, we use complete mathematical models.
Some of the methods are illustrated on the observations of the semi-regular pulsating variable RY UMa. The star shows a drastic incease of an amplitude of pulsations and a wide minimum of the mean brightness during the pulsation cycle.

Read this paper on arXiv…

I. Andronov and L. Chinarova
Tue, 22 Oct 19
83/91

Comments: Astronomical and Astrophysical Transactions (AApTr), 2019, v.32 (in press)