R/RcppExports.R

In g-l-mansell/RcppRidge: Parallel Bayesian Ridge Regression

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

#' Fit a single ridge regression model
#'
#' @param X Data matrix
#' @param y Column matrix of responses
#' @param lambda Numeric hyperparameter controlling the strength of the L2 penalisation (non-negative)
#' @return Vector of penalised regression coefficients
#' @export
fit_rr <- function(X, y, lambda) {
    .Call(`_RcppRidge_fit_rr`, X, y, lambda)
}

#' Predict new sample responses using a tuned regression model
#'
#' @param X Data matrix of test samples
#' @param beta Vector of regression coefficients
#' @return Vector of fitted values
#' @export
predict_rr <- function(X, beta) {
    .Call(`_RcppRidge_predict_rr`, X, beta)
}

#' Calculate leave one out cross validation error (OCV) for a single regression model
#'
#' @param X Data matrix
#' @param y Column matrix of responses
#' @param lambda Numeric hyperparameter controlling the strength of the L2 penalisation (non-negative)
#' @return Numeric OCV
#' @export
get_ocv_once <- function(X, y, lambda) {
    .Call(`_RcppRidge_get_ocv_once`, X, y, lambda)
}

#' Fast calculate leave one out cross validation error (OCV)
#'
#' Fast calculate OCV given a singular value decomposition (SVD) decomposition of data matrix X
#'
#' @param X Data matrix
#' @param y Column matrix of responses
#' @param lambda Numeric hyperparameter controlling the strength of the L2 penalisation (non-negative)
#' @param U Matrix U from SVD of X = UDV
#' @param s Elements of diagonal matrix D from SVD of X = UDV
#' @return Numeric OCV
#' @export
get_ocv <- function(X, y, lambda, U, s) {
    .Call(`_RcppRidge_get_ocv`, X, y, lambda, U, s)
}

#' Find the optimal regularisation parameter through optimised leave one out cross validation
#'
#' @param X Data matrix
#' @param y Column matrix of responses
#' @param lams Vector of regularisation parameters to test
#' @return Vector of OCVs
#' @export
optim_rr <- function(X, y, lams) {
    .Call(`_RcppRidge_optim_rr`, X, y, lams)
}

#' Fit a ridge regression model to multiple groups in parallel
#'
#' @param X Data matrix
#' @param y Column matrix of responses
#' @param lams Vector of regularisation parameters to test
#' @param idx Vector of sample groups 
#' @return List with two objects
#' lambdas A vector of the optimal value of lambda for each group
#' betas A matrix where columns are the fitted regression coefficients for each group 
#' variances A 3D array, where variances[,,1] will be the covariance matrix for B|y for the first model
#' @export
par_reg <- function(X, y, lams, idx) {
    .Call(`_RcppRidge_par_reg`, X, y, lams, idx)
}

#' Predict new samples using the results from par_reg
#'
#' @param X Data matrix of test samples
#' @param betas Matrix of regression coefficients
#' @param idx Vector of sample groups, corresponding to the columns of betas
#' (e.g. idx=c(1, 3) means betas[,1] will be used to predict X[1,], and betas[,3] will be used to predict X[2,])
#' @return Vector of fitted values
#' @export
predict_groups <- function(X, betas, idx) {
    .Call(`_RcppRidge_predict_groups`, X, betas, idx)
}

#' Sample from a multivariate Gaussian
#'
#' @param n Integer number of samples to draw
#' @param mu Vector of means
#' @param sigma Covariance matrix
#' @return Matrix of MVN samples (n rows)
#' @export
rmvn_omp <- function(n, mu, sigma) {
    .Call(`_RcppRidge_rmvn_omp`, n, mu, sigma)
}