Bayesian inference for cycles in noisy time series
Natural Sciences (Physics)
911 views
Date of upload:
15.11.2016
Co-author:
Abstract:
Assuming a simple data-generating model of the type "pure harmonics plus noise" and applying Bayesian inference techniques has two advantages over Fourier analysis: (i) lines that are separated by less than the Fourier resolution might be detected and (ii) the uncertainties of the detected lines can be quantified.
If the data stems from the solar activity, inferring the parameters of a stochastic dynamo model might be worthwhile.
