Implementations of a Bayesian parametric (autoregressive), a Bayesian nonparametric (Whittle likelihood with Bernstein-Dirichlet prior) and a Bayesian semiparametric (autoregressive likelihood with Bernstein-Dirichlet correction) procedure are provided
The work is based on the corrected parametric likelihood by Kirch, Meyer et al.
It was supported by DFG grant KI 1443/3-1.
See the examples and the referenced paper for further details.
The function gibbs_AR, gibbs_NP, gibbs_NPC correspond to the procedures AR, NP and NPC outlined in the simulation study in the referenced paper.
The other functions are useful utility functions.
C. Kirch et al. (2017) Beyond Whittle: Nonparametric Correction of a Parametric Likelihood With a Focus on Bayesian Time Series Analysis <arXiv:1701.04846>