Time Series Analysis of Non-Gaussian Observations Based on State Space Models from Both Classical and Bayesian Perspectives

Abstract

The analysis of non-Gaussian time series using state space models is considered from both classical and Bayesian perspectives. The treatment in both cases is based on simulation using importance sampling and antithetic variables; Markov chain Monte Carlo methods are not employed. Non-Gaussian disturbances for the state equation as well as for the observation equation are considered. Methods for estimating conditional and posterior means of functions of the state vector given the observations, and the mean-square errors of their estimates, are developed. These methods are extended to cover the estimation of conditional and posterior densities and distribution functions. Choice of importance sampling densities and antithetic variables is discussed. The techniques work well in practice and are computationally ef®cient. Their use is illustrated by applying them to a univariate discrete time series, a series with outliers and a volatility series.

Predicting Time-Varying Parameters with Parameter-Driven and Observation-Driven Models

“…Examples of specifications within this framework include stochastic volatility models as in Tauchen and Pitts (1983), Taylor (1986), Melino and Turnbull (1990) and Ghysels, Harvey, and Renault (1996), stochastic conditional duration models as in Bauwens and Veredas (2004), stochastic conditional intensity models as in Bauwens and Hautsch (2006), stochastic copulas as in Hafner and Manner (2011), and non-Gaussian unobserved components time series models as in Durbin and Koopman (2000).…”

Parameterisation and efficient MCMC estimation of non-Gaussian state space models

“…Adopting the terminology of Gelfand, Sahu and Carlin (1995), applied in the context of a random effects model, the model in (6) to (9) is referred to here as 'non-centred in location'. 3 From (9) and (10) it is clear that the location parameter, μ, now appears explicitly in the measurement equation.…”

“…Sampling μ and σ from their full conditional posterior distributions is achieved using the OBMC algorithm in Section 3.3, with ω = μ and ω = σ η respectively. 6 …”

Out of hours workload management: Bayesian inference for decision support in secondary care

“…On the other hand, observation-driven 255 specifications for Poisson autoregressive models offer straightforward and efficient ways to draw inference over short-horizons ( [24,38]). These benefit from easy to calculate likelihoods; however, stationarity and ergodicity properties are hard to derive, and they most importantly suffer from a lack of interpretability on covariates when compared to the parameter-driven alternative discussed in…”