bsad {bsamGP}R Documentation

Bayesian Semiparametric Density Estimation


This function fits a semiparametric model, which consists of parametric and nonparametric components, for estimating density using a logistic Gaussian process.


bsad(x, xmin, xmax, nint, MaxNCos, mcmc = list(), prior = list(),
smoother = c('geometric', 'algebraic'),
parametric = c('none', 'normal', 'gamma', 'laplace'), marginal.likelihood = TRUE,
verbose = FALSE)



a vector giving the data from which the density estimate is to be computed.


minimum value of x.


maximum value of x.


number of grid points for plots (need to be odd). The default is 201.


maximum number of Fourier coefficients.


a list giving the MCMC parameters. The list includes the following integers (with default values in parentheses): kappaloop (5) giving the number of MCMC loops within each choice of kappa, nblow (10000) giving the number of MCMC in transition period, nskip (10) giving the thinning interval, smcmc (1000) giving the number of MCMC for analysis, and ndisp (1000) giving the number of saved draws to be displayed on screen (the function reports on the screen when every ndisp iterations have been carried out).


a list giving the prior information. The list includes the following parameters (default values specify the non-informative prior): gmax giving maximum value for gamma (default = 5), PriorProbs giving prior probability of parametric and semiparametric models, beta_m0 and beta_v0 giving the hyperparameters for prior distribution of the parametric coefficients, r0 and s0 giving the hyperparameters of σ^2 for the logits, u0 and v0 giving the hyperparameters of τ^2 for Fourier coefficients, PriorKappa and KappaGrid giving prior on the number of cosine terms.


types of smoothing priors for Fourier coefficients. See Details.


specifying a distribution of the parametric part to be test.


a logical variable indicating whether the log marginal likelihood is calculated.


a logical variable. If TRUE, the iteration number and the Metropolis acceptance rate are printed to the screen.


This generic function fits a semiparametric model, which consists of parametric and nonparametric, for density estimation (Lenk, 2003):

f(x | β, Z) = \frac{\exp[h(x)^\topβ + Z(x)]}{\int_\mathcal{X} \exp[h(y)^\topβ + Z(y)]dG(y)}

where Z is a zero mean, second-order Gaussian process with bounded, continuous covariance function. i.e.,

E[Z(x), Z(y)] = σ(x,y), \quad \int_\mathcal{X}ZdG = 0 ~~(a.s.)

Using the Karhunen-Loeve Expansion, Z is represented as infinite series with random coefficients

Z(x) = ∑_{j=1}^∞ θ_j\varphi_j(x),

where \{\varphi_j\} is the cosine basis, \varphi_j(x)=√{2}\cos[jπ G(x)].

For the random Fourier coefficients of the expansion, two smoother priors are assumed (optional),

θ_j | τ, γ \sim N(0, τ^2\exp[-jγ]), ~ j ≥ 1 ~ (geometric ~smoother)

θ_j | τ, γ \sim N(0, τ^2\exp[-ln(j+1)γ]), ~ j ≥ 1 ~ (algebraic ~smoother)

The coefficient β have the popular normal prior,

β | m_{0,β}, V_{0,β} \sim N(m_{0,β}, V_{0,β})

To complete the model specification, independent hyper priors are assumed,

τ^2 | r_0, s_0 \sim IGa(r_0/2, s_0/2)

γ | w_0 \sim Exp(w_0)

Note that the posterior algorithm is based on computing a discrete version of the likelihood over a fine mesh on \mathcal{X}.


An object of class bsad representing the Bayesian spectral analysis density estimation model fit. Generic functions such as print, fitted and plot have methods to show the results of the fit.

The MCMC samples of the parameters in the model are stored in the list mcmc.draws, the posterior samples of the fitted values are stored in the list fit.draws, and the MCMC samples for the log marginal likelihood are saved in the list loglik.draws. The output list also includes the following objects:


posterior estimates for all parameters in the model.


log marginal likelihood.


posterior probability of models.


the matched call.


running time of Markov chain from system.time().


Jo, S., Choi, T., Park, B. and Lenk, P. (2019). bsamGP: An R Package for Bayesian Spectral Analysis Models Using Gaussian Process Priors. Journal of Statistical Software, 90, 310-320.

Lenk, P. (2003) Bayesian semiparametric density estimation and model verification using a logistic Gaussian process. Journal of Computational and Graphical Statistics, 12, 548-565.


## Not run: 
# Old Faithful geyser data #

# mcmc parameters
mcmc <- list(nblow = 10000,
	           smcmc = 1000,
	           nskip = 10,
	           ndisp = 1000,
	           kappaloop = 5)

# fits BSAD model
fout <- bsad(x = eruptions, xmin = 0, xmax = 8, nint = 501, mcmc = mcmc,
             smoother = 'geometric', parametric = 'gamma')

# Summary
print(fout); summary(fout)

# fitted values
fit <- fitted(fout)

# predictive density plot
plot(fit, ask = TRUE)


## End(Not run)

[Package bsamGP version 1.2.3 Index]