AutoNOM
is a Julia software to model nonstationary
periodic time series. This Bayesian method approximates the time series using a piecewise oscillatory model with unknown periodicities, to estimate the change-points while simultaneously identifying the potentially changing periodicities in the data. Inference is carried out through reversible-jump MCMC based
algorithms as detailed in Hadj-Amar et al. (2020) "Bayesian Model Search for Nonstationary Periodic Time Series", published in JASA. The software is implemented in Julia 1.6
.
We provide a snapshot of illustrative_example.jl
, which contains a tutorial for using our software in Julia
- Run RJMCMC sampler
MCMC_simul = AutoNOM(data, hyperparms; s_start = [40])