est_quancov {TLMoments}

Estimate the covariance matrix of quantile estimations

Description

Internal function. Use est_cov. Description not done yet.

Usage

est_quancov(x, distr = "", p = NULL, leftrim = 0L, rightrim = 0L, ...)

## S3 method for class 'numeric'
est_quancov(x, distr, p, leftrim = 0L, rightrim = 0L, np.cov = FALSE, ...)

## S3 method for class 'matrix'
est_quancov(
  x,
  distr,
  p,
  leftrim = 0L,
  rightrim = 0L,
  np.cov = FALSE,
  reg.weights = NULL,
  ...
)

## S3 method for class 'quantiles'
est_quancov(
  x,
  distr = attr(x, "distribution"),
  p = attr(x, "p"),
  leftrim = attr(x, "source")$trimmings[1],
  rightrim = attr(x, "source")$trimmings[2],
  set.n = NA,
  ...
)

Arguments

x

numeric vector or matrix containing data.

distr

character of length 1 giving the distribution if parametric assumption should be used.

p

quantile levels from which the covariance should be calculated.

leftrim, rightrim

lower and upper trimming parameter used for parameter calculation, have to be non-negative integers.

...

additional arguments.

np.cov

boolean, if TRUE no parametric assumptions are used to calculate the covariance matrix (default FALSE).

reg.weights

numeric vector of weights for regionalized TLMoments.

set.n

hypothetical data length n if theoretical values are given.

Value

numeric matrix

Examples

### Numeric vectors
x <- rgev(500, shape = .2)

quantiles(parameters(TLMoments(x), "gev"), c(.9, .95, .99))
est_quancov(x, "gev", c(.9, .95, .99), 0, 0)
#cov(t(replicate(5000,
#  quantiles(parameters(TLMoments(rgev(500, shape = .2)), "gev"), c(.9, .95, .99))
#)))

quantiles(parameters(TLMoments(x, rightrim = 1), "gev"), c(.9, .95, .99))
est_quancov(x, "gev", c(.9, .95, .99), 0, 1)
#cov(t(replicate(5000,
#  quantiles(
#    parameters(TLMoments(rgev(500, shape = .2), rightrim = 1), "gev"),
#    c(.9, .95, .99)
#  )
#)))

### Numeric matrices
x <- matrix(rgev(600, shape = .2), nc = 3)

quantiles(parameters(TLMoments(x), "gev"), c(.9, .95, .99))
est_quancov(x, "gev", c(.9, .95, .99), 0, 0)

est_quancov(x, "gev", .9, 0, 0)
#cov(t(replicate(5000,
# quantiles(
#   parameters(TLMoments(matrix(rgev(600, shape = .2), nc = 3)),
#  "gev"), .9)
#  )
#))

### quantiles object
q <- quantiles(as.parameters(loc = 3, scale = 2, shape = .4, distr = "gev"), c(.9, .99))
est_quancov(q)
est_quancov(q, leftrim = 0, rightrim = 0)
est_quancov(q, leftrim = 0, rightrim = 0, set.n = 10)