R/RcppExports.R

In xhuang20/rdcqr: Local Composite Quantile Regression Method for Regression Discontinuity

# Generated by using Rcpp::compileAttributes() -> do not edit by hand
# Generator token: 10BE3573-1514-4C36-9D1C-5A225CD40393

#' @title Function to compute the local composite quantile regression estimate
#' 
#' @description This function computes the local composite quantile regression 
#' estimate. The point of interest can be either interior or boundary. 
#' 
#' @param x0 point of interest
#' @param x_vec a vector of covariates
#' @param y a vector of dependent variable, the treatment outcome variable 
#' in the case of regression discontinuity.
#' @param kernID kernel ID for different kernels.
#' \enumerate{
#'   \item \code{kernID = 0}: triangular kernel.
#'   \item \code{kernID = 1}: biweight kernel.
#'   \item \code{kernID = 2}: Epanechnikov kernel. 
#'   \item \code{kernID = 3}: Gaussian kernel.
#'   \item \code{kernID = 4}: tricube kernel.
#'   \item \code{kernID = 5}: triweight kernel. 
#'   \item \code{kernID = 6}: uniform kernel. 
#' }
#' @param tau A vector of quantile positions. They are obtained by 
#' \code{tau = (1:q)/(q+1)}.
#' @param h A scalar bandwidth.
#' @param p The polynomial degree.
#' @param maxit Maximum iteration number in the MM algorithm for quantile
#'  estimation. Defaults to 100.
#' @param tol Convergence criterion in the MM algorithm. Defaults to 1e-4.
#' @export
#' @return \code{cqrMMcpp} returns a list with the following components:
#' \item{beta0}{A q by 1 vector of estimates for q quantiles.}
#' \item{beta1}{A p by 1 vector of estimates for the first p derivatives of
#' the conditional mean function.}
#' @examples
#' # Use the Head Start data as an example.
#' data(headstart)
#' data_n = subset(headstart, headstart$poverty < 0)
#' q      = 5
#' tau    = (1:q) / (q + 1)
#' 
#' # Compute the local composite quantile estimate
#' est = cqrMMcpp(x0     = 0,
#'                x_vec  = data_n$poverty,
#'                y      = data_n$mortality,
#'                kernID = 2,
#'                tau    = tau,
#'                h      = 4.0,
#'                p      = 1,
#'                maxit  = 100,
#'                tol    = 1.0e-4)
#'                
#' # Estimate of the conditional mean on the boundary
#' est_mean = mean(est$beta0)
#' 
#' # Estimate of the first derivative of the conditional mean function
#' est_d1   = est$beta1[1]
cqrMMcpp <- function(x0, x_vec, y, kernID, tau, h, p, maxit = 100L, tol = 1e-4) {
    .Call(`_rdcqr_cqrMMcpp`, x0, x_vec, y, kernID, tau, h, p, maxit, tol)
}

#' @title Function to compute the fitted values and residuals
#' 
#' @description This function computes the fitted values and residuals 
#' in a local composite quantile regression.
#' 
#' @param x_vec A vector of covariates.
#' @param y a vector of dependent variable, the treatment outcome variable
#' in the case of regression discontinuity.
#' @param kernID The kernel id that includes
#' \enumerate{
#'   \item \code{kernID = 0}: triangular kernel.
#'   \item \code{kernID = 1}: biweight kernel.
#'   \item \code{kernID = 2}: Epanechnikov kernel. 
#'   \item \code{kernID = 3}: Gaussian kernel.
#'   \item \code{kernID = 4}: tricube kernel.
#'   \item \code{kernID = 5}: triweight kernel. 
#'   \item \code{kernID = 6}: uniform kernel. 
#' }
#' @param tau A vector of quantile positions. They are obtained by 
#' \code{tau = (1:q)/(q+1)}.
#' @param h A scalar bandwidth.
#' @param p The polynomial degree. Defaults to 1.
#' @param maxit Maximum number of iterations in the MM algorithm. Defaults to 100.
#' @param tol The convergence criterion. Defaults to 1.0e-4.
#' @param parallel Set it to 1 if using parallel computing. Default is 1.
#' @param grainsize The minimum chunk size for parallelization. Defaults to 1.
#' @export
#' @return \code{est_cqr} returns a list with the following components:
#' \item{y_hat}{The fitted value at each point of the input vector \code{x_vec}.}
#' \item{u_hat}{The residual vector.}
#' \item{sig_hat}{The estimated standard deviation at each point of the input 
#' vector \code{x_vec}.}
#' \item{e_hat}{The scaled residual vector. It is the residual vector divided 
#' by the \code{sig_hat} vector.}
#' @examples
#' # Use the Head Start data as an example.
#' data(headstart)
#' data_n = subset(headstart, headstart$poverty < 0)
#' q      = 5
#' tau    = (1:q) / (q + 1)
#' est_cqr(x_vec  = data_n$poverty,
#'         y      = data_n$mortality,
#'         kernID = 2,
#'         tau    = tau,
#'         h      = 4.0,
#'         p      = 1) 
est_cqr <- function(x_vec, y, kernID, tau, h, p = 1L, maxit = 100L, tol = 1e-4, parallel = 0L, grainsize = 1L) {
    .Call(`_rdcqr_est_cqr`, x_vec, y, kernID, tau, h, p, maxit, tol, parallel, grainsize)
}

#' @title Function to compute the local composite quantile regression estimate with covariates
#' 
#' @description This function computes the local composite quantile regression 
#' estimate with covariates. The point of interest can be either interior or boundary. 
#' 
#' @param x0 point of interest
#' @param x_vec a vector of covariates
#' @param y a vector of dependent variable, the treatment outcome variable 
#' in the case of regression discontinuity.
#' @param xcv A matrix of additional covariates to be used in local regression analysis.
#' It must have the same number of rows as the running variable \code{x}. When calling
#' the function \code{cqrMMxcv} directly, if there is no additional covariate, set 
#' xcv as an arbitrary zero matrix, e.g., \code{xcv = matrix(0,2,2)}, as the corresponding
#' function argument value.
#' @param kernID kernel ID for different kernels.
#' \enumerate{
#'   \item \code{kernID = 0}: triangular kernel.
#'   \item \code{kernID = 1}: biweight kernel.
#'   \item \code{kernID = 2}: Epanechnikov kernel. 
#'   \item \code{kernID = 3}: Gaussian kernel.
#'   \item \code{kernID = 4}: tricube kernel.
#'   \item \code{kernID = 5}: triweight kernel. 
#'   \item \code{kernID = 6}: uniform kernel. 
#' }
#' @param tau A vector of quantile positions. They are obtained by 
#' \code{tau = (1:q)/(q+1)}.
#' @param h A scalar bandwidth.
#' @param p The polynomial degree.
#' @param maxit Maximum iteration number in the MM algorithm for quantile
#'  estimation.
#' @param tol Convergence criterion in the MM algorithm.
#' @export
#' @return \code{cqrMMxcv} returns a list with the following components:
#' \item{beta0}{A q by 1 vector of estimates for q quantiles.}
#' \item{beta1}{A p by 1 vector of estimates for the first p derivatives of
#' the conditional mean function.}
#' @examples
#' # Use the Head Start data as an example.
#' data(headstart)
#' data_n = subset(headstart, headstart$poverty < 0)
#' q      = 5
#' tau    = (1:q) / (q + 1)
#' 
#' # Compute the local composite quantile estimate
#' est = cqrMMxcv(x0     = 0,
#'                x_vec  = data_n$poverty,
#'                y      = data_n$mortality,
#'                xcv    = matrix(0,2,2),
#'                kernID = 2,
#'                tau    = tau,
#'                h      = 4.0,
#'                p      = 1,
#'                maxit  = 10,
#'                tol    = 1.0e-3)
#'                
#' # Estimate of the conditional mean on the boundary
#' est_mean = mean(est$beta0)
#' 
#' # Estimate of the first derivative of the conditional mean function
#' est_d1   = est$beta1[1]
cqrMMxcv <- function(x0, x_vec, y, xcv, kernID, tau, h, p, maxit = 100L, tol = 1e-4) {
    .Call(`_rdcqr_cqrMMxcv`, x0, x_vec, y, xcv, kernID, tau, h, p, maxit, tol)
}