NGSSEML: Non-Gaussian State Space with Exact Marginal Likelihood

The number of packages/software for Gaussian State Space models has increased over recent decades. However, there are very few codes available for non-Gaussian State Space (NGSS) models due to analytical intractability that prevents exact calculations. One of the few tractable exceptions is the family of NGSS with exact marginal likelihood, named NGSSEML. In this work, we present the wide range of data formats and distributions handled by NGSSEML and a package in the R language to perform classical and Bayesian inference for them. Special functions for filtering, forecasting, and smoothing procedures and the exact calculation of the marginal likelihood function are provided. The methods implemented in the package are illustrated for count and volatility time series and some reliability/survival models, showing that the codes are easy to handle. Therefore, the NGSSEML family emerges as a simple and interesting option/alternative for modeling non-Gaussian time-varying structures commonly encountered in time series and reliability/survival studies. Keywords: Bayesian, classical inference, reliability, smoothing, time series, software R

Thiago R. Santos , Glaura C. Franco , Dani Gamerman

CRAN packages used

StructTS, dlm, dlmodeler, SSsimple, MARSS, sspir, pomp, KFAS, bssm, dynamichazard, NGSSEML, coda

CRAN Task Views implied by cited packages

TimeSeries, Bayesian, DifferentialEquations, Finance, GraphicalModels


Text and figures are licensed under Creative Commons Attribution CC BY 4.0. The figures that have been reused from other sources don't fall under this license and can be recognized by a note in their caption: "Figure from ...".


For attribution, please cite this work as

Santos, et al., "The R Journal: NGSSEML: Non-Gaussian State Space with Exact Marginal Likelihood", The R Journal, 2021

BibTeX citation

  author = {Santos, Thiago R. and Franco, Glaura C. and Gamerman, Dani},
  title = {The R Journal: NGSSEML: Non-Gaussian State Space with Exact Marginal Likelihood},
  journal = {The R Journal},
  year = {2021},
  note = {},
  doi = {10.32614/RJ-2021-087},
  volume = {13},
  issue = {2},
  issn = {2073-4859},
  pages = {208-227}