bbeed {ExtremalDep}  R Documentation 
Nonparametric Bayesian estimation of the extremal dependence function (angular measure and Pickands dependence function) for twodimensional data on the basis of Bernstein polynomials representations.
bbeed(data, pm0, param0, k0, hyperparam, nsim, nk = 70,
prior.k = c('pois','nbinom'), prior.pm = c('unif','beta', 'null'),
lik = TRUE)
data 

pm0 
list; the initial state of the Markov chain of point masses 
param0 
real vector in (0,1) of length 
k0 
postive integer indicating the initial state of the Markov chain of the Bernstein polynomial's degree. Randomly generated if missing. 
hyperparam 
list containing the hyperparameters (see Details). 
nsim 
postive integer indicating the number of the iterations of the chain. 
nk 
positive integer, maximum value of 
prior.k 
string, indicating the prior distribution on the degree of the Bernstein polynomial, 
prior.pm 
string, indicating the prior distribution on the point masses 
lik 
logical; if 
The hyperparameters and starting values may not be specified and in this case a warning message will be printed and default values are set. In particular if hyperparam
is missing we have
hyperparam < NULL
hyperparam$a.unif < 0
hyperparam$b.unif < .5
hyperparam$a.beta < c(.8,.8)
hyperparam$b.beta < c(5,5)
mu.pois < hyperparam$mu.pois < 4
mu.nbinom < hyperparam$mu.nbinom < 4
var.nbinom < 8
pnb < hyperparam$pnb < mu.nbinom/var.nbinom
rnb < hyperparam$rnb < mu.nbinom^2 / (var.nbinommu.nbinom)
The routine returns an estimate of the Pickands dependence function using the Bernstein polynomials approximation proposed in Marcon et al. (2016).
The method is based on a preliminary empirical estimate of the Pickands dependence function.
If you do not provide such an estimate, this is computed by the routine. In this case, you can select one of the empirical methods
available. est = "ht"
refers to the HallTajvidi estimator (Hall and Tajvidi 2000).
With est = "cfg"
the method proposed by Caperaa et al. (1997) is considered. Note that in the multivariate case the adjusted version of Gudendorf and Segers (2011) is used. Finally, with est = "md"
the estimate is based on the madogram defined in Marcon et al. (2016).
For each row of the (m \times d)
design matrix x
is a point in the unit d
dimensional simplex,
S_d := \left\{ (w_1,\ldots, w_d) \in [0,1]^{d}: \sum_{i=1}^{d} w_i = 1 \right\}.
With this "regularization" method, the final estimate satisfies the neccessary conditions in order to be a Pickands dependence function.
A(\bold{w}) = \sum_{\bold{\alpha} \in \Gamma_k} \beta_{\bold{\alpha}} b_{\bold{\alpha}} (\bold{w};k).
The estimates are obtained by solving an optimization quadratic problem subject to the constraints. The latter are represented by the following conditions:
A(e_i)=1; \max(w_i)\leq A(w) \leq 1; \forall i=1,\ldots,d;
(convexity).
The order of polynomial k
controls the smoothness of the estimate. The higher k
is, the smoother the final estimate is.
Higher values are better with strong dependence (e. g. k=23
), whereas small values (e.g. k=6
or k=10
) are enough with mild or weak dependence.
An empirical transformation of the marginals is performed when margin="emp"
. A maxlikelihood fitting of the GEV distributions is implemented when margin="est"
. Otherwise it refers to marginal parametric GEV theorethical distributions (margin = "exp", "frechet", "gumbel"
).
If bounds = TRUE
, a modification can be implemented to satisfy
\max(w,1w) \leq A_n(w) \leq 1 \forall 0 \leq w \leq 1
:
A_n(w) = \min(1, \max{A_n(w), w, 1w}).
return(list(pm=spm, eta=seta, k=sk, accepted=accepted/nsim, prior = list(hyperparam=hyperparam, k=prior.k, pm=prior.pm)))
beta 
vector of polynomial coefficients 
A 
numeric vector of the estimated Pickands dependence function 
Anonconvex 
preliminary nonconvex function 
extind 
extremal index 
The number of coefficients depends on both the order of polynomial k
and the dimension d
. The number of parameters is d1 \choose k+d1
.
The size of the vector beta
must be compatible with the polynomial order k
chosen.
With the estimated polynomial coefficients, the extremal coefficient, i.e. d*A(1/d,\ldots,1/d)
is computed.
Simone Padoan, simone.padoan@unibocconi.it, https://mypage.unibocconi.it/simonepadoan/; Boris Beranger, borisberanger@gmail.com https://www.borisberanger.com/; Giulia Marcon, giuliamarcongm@gmail.com
Marcon G., Padoan, S.A. and AntonianoVillalobos I. (2016) Bayesian Inference for the Extremal Dependence. Electronic Journal of Statistics, 10(2), 33103337.
if (interactive()){
# This reproduces some of the results showed in Fig. 1 (Marcon, 2016).
set.seed(1890)
data < evd::rbvevd(n=100, dep=.6, asy=c(0.8,0.3), model="alog", mar1=c(1,1,1))
nsim = 500000
burn = 400000
mu.nbinom = 3.2
var.nbinom = 4.48
hyperparam < list(a.unif=0, b.unif=.5, mu.nbinom=mu.nbinom, var.nbinom=var.nbinom)
k0 = 5
pm0 = list(p0=0.06573614, p1=0.3752118)
eta0 = ExtremalDep:::rcoef(k0, pm0)
mcmc < bbeed(data, pm0, eta0, k0, hyperparam, nsim,
prior.k = "nbinom", prior.pm = "unif")
w < seq(0.001, .999, length=100)
summary.mcmc < summary.bbeed(w, mcmc, burn, nsim, plot=TRUE)
}