Simultaneous modeling of the quantile and the expected shortfall of a response variable given
a set of covariates, see Dimitriadis and Bayer (2019)
The goal of esreg is to simultaneously model the quantile and the Expected Shortfall of a response variable given a set of covariates.
You can install the released version from CRAN:
The latest version of the package is under development at GitHub. You can install the development version using these commands:
If you are using Windows, you need to install the Rtools for compilation of the codes.
# Load the esreg package library(esreg) # Simulate data from DGP-(2) in the paper set.seed(1) x <- rchisq(1000, df = 1) y <- -x + (1 + 0.5 * x) * rnorm(1000) # Estimate the model and the covariance fit <- esreg(y ~ x, alpha = 0.025) cov <- vcov(object = fit, sparsity = "nid", cond_var = "scl_sp")
Improved speed of the semi-parametric covariance estimator
Fixed an overloaded ‘pow(int&, int)’ bug (Solaris), improved the help files and marked several functions as internal.
Bump version for CRAN release
Clean imports in description
Add the residuals methods
Add the semi-parametric estimator of the truncated conditional variance
Replace the random restart optimizer with the iterated local search
Move GenSA to the optional packages
Add estimation of the truncated conditional variance using the skewed Student-t distribution.
Remove the one_shot estimation method: use random_restart instead.
Added an estimator of the asymptotic covariance of the two-step estimator
Added a (extremely fast but less precise) two-step estimator
Added new specification functions
Added the Z-estimator